Traumatic brain injury in mice generates early-stage Alzheimer’s disease related protein pathology that correlates with neurobehavioral deficits

Traumatic brain injury (TBI) increases the long-term risk of neurodegenerative diseases, including Alzheimer’s disease (AD). Here, we demonstrate that protein variant pathology generated in brain tissue of an experimental TBI mouse model is similar to protein variant pathology observed in human ADbrains, and that subacute accumulation of two AD associated variants of amyloid beta (Aβ) and tau in the TBI mouse model correlated with behavioral deficits. Male C57BL/6 mice were subjected to midline fluid percussion injury or to sham injury, after which sensorimotor function (rotarod, neurological severity score), cognitive deficit (novel object recognition), and affective deficits (elevated plus maze, forced swim task) were assessed at different days post-injury (DPI). Protein pathology at 7, 14, and 28 DPI was measured in multiple brain regions using an immunostain panel of reagents selectively targeting different neurodegenerative disease-related variants of Aβ, tau, TDP-43, and alpha-synuclein. Overall, TBI resulted in sensorimotor deficits and accumulation of AD-related protein variant pathology near the impact site, both of which returned to sham levels by 14 DPI. Individual mice, however, showed persistent behavioral deficits and/or accumulation of selected toxic protein variants at 28 DPI. Behavioral outcomes of each mouse were correlated with levels of seven different protein variants in ten brain regions at specific DPI. Out of 21 significant correlations between protein variant levels and behavioral deficits, 18 were with variants of Aβ or tau. Correlations at 28 DPI were all between a single Aβ or tau variant, both of which are strongly associated with human AD cases. These data provide a direct mechanistic link between protein pathology resulting from TBI and the hallmarks of AD.


Introduction
Traumatic brain injury (TBI) affects over two million people each year, and is now the principal cause of death among children and young adults [16,48]. Even though there is usually an initial resolution of physical symptoms following TBI, long-term cellular damage in the brain can occur [1] leading to an increased susceptibility for a spectrum of neurodegenerative disorders [13]. Neurons are particularly vulnerable to the high sheer forces and mechanical deformation induced by TBI [4,25,26,36,47], where the resulting damage to protein transport mechanisms can result in axonal accumulation of neuro lament proteins, such as tau [47]. Aggregates of tau are also the major component of the hallmark neuro brillary tangles in Alzheimer's disease (AD) brain. In addition to changes in tau distribution, cellular stress induced by TBI can lead to a cascade of in ammatory processes in the brain and an increase in neuronal expression of both the amyloid precursor protein (APP) and beta-site APP cleaving enzyme (BACE-1) [2,27,49]. The increased expression of APP and BACE-1 increases amyloid-beta (Aβ) levels which can lead to toxic aggregation and accumulation of Aβ, also a characteristic feature of AD.
The location, severity, and number of brain injuries can affect the evolution of different neurodegenerative diseases since the different pathogeneses involved may be attributed to the type and local environment of the neuronal cells damaged and ultimately the unique composition of toxic protein variants generated [5,16,32]. Here, we hypothesized that experimental TBI would generate different toxic protein ngerprints of tau, Aβ, TDP-43, and a-syn variants in individual mice, and that speci c protein variants would correlate with injury-induced behavioral de cits. In particular, both the spatial and temporal pro les of the toxic variants that accumulate in the brain may help determine how the TBI associated neuronal stress contributes to long-term neurodegenerative disorders.
We previously developed a panel of single chain antibody variable domain (scFv) based reagents that very selectively bind disease-related variants of tau, Aβ, TDP-43, and α-syn that are present in postmortem validated human neurodegenerative disease brain tissue, but not in cognitively normal, agematched human samples [11,12,30,50,53,61]. We have shown that this panel of reagents can be used to identify protein variant ngerprints that correlate with different neurodegenerative diseases, including AD, PD, FTD, and ALS [51-54, 56, 58]. We have also shown that these same protein variants are detected in the chronic TBI human brain, where respective protein variant ngerprints of blood samples taken from individual TBI cases mirror the protein variant ngerprint signature observed for speci c neurodegenerative diseases [55]. Here, we used a panel of these neurodegenerative disease-selective reagents to histologically characterize brain tissue of mice subjected to diffuse TBI. We determined the extent of staining of each protein variant both to speci c brain regions and as a function of time postinjury, and correlated individual anatomical protein variant staining levels with neurobehavioral performance.

Rigor
All animal studies were conducted in accordance with the guidelines established by the internal IACUC (Institutional Animal Care and Use Committee) and the NIH guidelines for the care and use of laboratory animals. Studies are reported following the Animal Research: Reporting In Vivo Experiments (ARRIVE) guidelines. Randomization of animals was achieved by assigning individuals to treatment groups prior to initiation of the study to ensure equal distribution among groups. Data collection stopped at predetermined nal endpoints based on days post-injury (DPI). Animals were excluded from the study if post-operative weight decreased by ≥ 15% of pre-surgical weight or baseline rotarod score was not met. No animals were excluded based on these criteria. All animal behavior and histology were scored by investigators blinded to the treatment groups. Tissue samples analyzed in this study were generated from animals used in a previous study that non-invasively examined acute sleep after TBI [40].

Animals
Male C57BL/6 mice (Harlan Laboratories, Inc., Indianapolis, IN) were used for all experiments. Mice were housed in a 14h light:10h dark cycle at a constant temperature (23°C ± 2° C) with food and water available ad libitum. After surgery, mice were evaluated daily during post-operative care via physical examination and documentation of each animal's condition. Animal care was approved by the Institutional Animal Care and Use Committee at the University of Arizona (13-460).

Midline uid percussion injury (mFPI)
Adult male mice (20-24g; n = 52) were subjected to midline uid percussion injury (mFPI) consistent with methods previously described [22,35,[41][42][43]. Mice were anesthetized using 5% iso urane in 100% oxygen for ve minutes and the head of the mouse was placed in a stereotaxic frame with continuously delivered iso urane at 2.5% via nosecone. While anesthetized, body temperature was maintained using a Deltaphase® isothermal heating pad (Braintree Scienti c Inc., Braintree, MA). A midline incision was made exposing bregma and lambda, and fascia was removed from the surface of the skull. A trephine (3 mm outer diameter) was used for the craniectomy, centered on the sagittal suture between bregma and lambda without disruption of the dura. An injury cap prepared from the female portion of a Luer-Loc needle hub was xed over the craniotomy using cyanoacrylate gel and methyl-methacrylate (Hygenic Corp., Akron, OH). The incision was sutured at the anterior and posterior edges and topical Lidocaine ointment was applied. The injury hub was closed using a Luer-Loc cap and mice were placed in a heated recovery cage and monitored until they were ambulatory before being returned to their home cages.
For injury induction 24 hours post-surgery, mice were re-anesthetized with 5% iso urane delivered for three minutes. The cap was removed from the injury-hub assembly and the dura was visually inspected through the hub to ensure it was intact with no debris. The hub was then lled with normal saline and attached to an extension tube connected to the male end of the uid percussion device (Custom Design and Fabrication, Virginia Commonwealth University, Richmond, VA). An injury of moderate severity (1.4 atm) was administered by releasing the pendulum onto the uid-lled cylinder following the return of a toe-pinch response. Sham-injured mice underwent the same procedure except the pendulum was not released. The injury hub was removed, and the brain was inspected for uniform herniation and integrity of the dura. The incision was cleaned with saline and closed using sutures. Mice received either sham injury, a single TBI, or two TBIs (3-9 hour injury interval). For mice that received two TBIs, the injury hub was left intact following the rst TBI and the mouse was reattached to the injury device for a second TBI before the hub was removed, the dura was inspected, and the incision was closed. To control for the anesthetic used during the induction of injury, uninjured shams and the single TBI group also received a second exposure to iso urane. After spontaneously righting, mice were placed in a heated recovery cage and monitored until they were ambulatory before being returned to their cage. Adequate measures were taken to minimize pain or discomfort.

Rotarod
Sensorimotor function was assessed using a rotarod (Economex Rotarod, Columbus Instruments, Columbus, OH). Mice were acclimated to the apparatus and trained for three days prior to the craniectomy surgery. During the acclimation phase, mice were placed on the stationary rod and allowed to explore for 30 seconds. The mice were trained by placing them on the rod at a constant speed of ve revolutions per minute (rpm) until the mice could walk 15 seconds at ve rpm. Next, mice were placed on the rod with an initial rotation speed of ve rpm and an acceleration of 0.2 rpm/sec. The trial ended when the mouse fell off the rod; the acclimation period ended after two trials. Mice were trained over three consecutive days and the last training session was recorded as a baseline score. Testing occurred at 2, 5, and 7 DPI. For the training and testing phase, two trials were run back-to-back and mice were returned to cages thereafter. After 10 minutes, mice had a third trial. Time spent on the rotarod from the best two trials were averaged to generate a time score for each mouse.
Neurological Severity Score (NSS) Sensorimotor function and balance were assessed using a modi ed neurological severity score at 2, 5, and 7 DPI. Mice were observed for hind limb exion, startle re ex, and seeking behavior (presence of these behaviors was considered successful task completion). Mice traversed in sequence, three, two, and one-centimeter beams. The beams were elevated, and mice were given one minute to travel 30 centimeters. The task was scored as a success if the mouse traveled 30 centimeters without a forelimb/hindlimb hanging from the beam. Mice were also required to balance on a 0.5-centimeter beam and a 0.5-centimeter round rod for three seconds in a stationary perched position (front paws between hind paws). One point was given for failure on an individual task, whereas no points were given if a mouse completed a task successfully. A composite score was assigned to each mouse that ranged from zero to eight. A high composite NSS score is indicative of task failure and interpreted as neurological impairment.

Elevated Plus Maze (EPM)
Anxiety-like behavior was assessed using the elevated plus maze. Mice were placed at the junction of the four arms, facing an open arm, and given ve minutes to explore the apparatus. The elevated plus maze consisted of two open arms and two arms with high walls. Mice were visually recorded using an overhead camera and recordings were analyzed using EthoVision XT 10 software. Time spent in the open arms and the closed arms were calculated. Open arm activity was interpreted as anti-anxiety behavior.

Forced swim task (FST)
Depressive-like behavior was assessed using the FST. Mice were placed into glass cylinders (15 cm diameter ×24.5 cm high) lled with water (25º C) for six minutes. The rst minute was excluded from analysis as an acclimation phase. Mice were visually recorded using an overhead camera and recordings were analyzed using EthoVision XT 10 software. Time spent actively swimming and time spent immobile was calculated for each mouse.

Novel Object Recognition (NOR)
Cognitive impairment was tested using the NOR test as previously published [21,42]. The test consisted of three phases: habituation, training, and testing. On the day of the test, mice were placed in an open eld (42 × 21 × 21 cm) for one hour of habituation. Mice were removed from the open eld and two identical objects were placed in opposing corners for the training phase. Mice were placed in the center of the eld and given ve minutes to explore the objects. Following training, mice were returned to their home cages for four hours. For the testing phase, one familiar object was placed in the original location and one novel object was placed in the opposing corner. Mice were placed into the center and given ve minutes to explore. For testing, the times spent actively investigating the novel and familiar object were quanti ed. Investigation of an object included the mice sni ng, touching, or climbing onto an object while facing the object. A discrimination index (DI) was calculated for each mouse: DI = × 100.

Tissue preparation
Following behavioral assessment, brain tissue was collected and brain pathology was characterized as a function of TBI and time post-injury. Brain tissue was harvested at 7, 14, or 28 DPI. Mice were given an overdose of sodium pentobarbital intraperitoneally and transcardially perfused with a phosphate buffered saline (PBS) ush followed by 4% paraformaldehyde. The brains were transferred to fresh xative solution and sent to NeuroScience Associates Inc. (Knoxville, TN) to be processed and sectioned. Brains were embedded into a gelatin matrix where they could be frozen and sectioned from one solid block (MultiBrain® Technology, NeuroScience Associates). Sections of 40 µm thickness were taken in the coronal plane and returned to the laboratory for free oating staining.

ScFv-Phage production
The seven different scFvs utilized (Table I) [11,12,30,50,61] were each expressed on the surface on bacteriophage as described [59]. Brie y, the expression plasmids encode the scFv connected to the M13 phage minor coat protein pIII, for expression of the scFv on the phage surface. TG1 cells containing the plasmids were incubated with M13 hyperphage (Progen, Germany) or KM13 helper phage overnight to generate the scFv-phage fusion construct. Scfv fusion phage were puri ed by repeated polyethylene glycol (PEG 8000) precipitation and centrifugation.

Chromogenic DAB Immunostaining
Brain tissue was stained for the presence of different variants of Aβ, tau, TDP-43, and α-syn utilizing a panel of seven different scFvs [11,12,30,50,61], where each scFv binds a different neurodegenerative disease related protein variant (Table 1). Brain sections of each mouse were separately probed with each of the seven scFvs and staining intensities of 10 different brain regions were determined (Fig. 1A).
Immunostaining was performed as described previously [23,24]. Tissue sections underwent a 10-minutehigh-temperature antigen retrieval step and incubated with scFv-phage (1:1000) for one hour; then placed in anti-M13 mouse antibody (Invitrogen 1:4000) incubation for one hour. Sections were washed and incubated with biotinylated anti-mouse IgG horse antibody (1:1000). Following washing, the Vectastain ABC-HRP kit (Vector Laboratories, Burlingame, CA) was applied. Samples were visualized using 3,3′diaminobenzidine (DAB) as a substrate (Vector Laboratories). Sections were counterstained with hematoxylin. Since all 39 mice brain slices were embedded into a single MultiBrain slice, staining intensities of each brain slice could be compared.

Immunostaining Image Analysis
Bright eld images were collected at 4x, 10x, and 40x using a Keyence BZ-X800 and a Leica TCS SP5 LSCM or EVOS M7000 in the Regenerative Medicine Imaging Facility at ASU. 4x image stitching was performed using Keyence software or NIH Image J 1.52p (FIJI 2). Antibody staining was measured using NIH ImageJ with a measured mean grey value intensity (integrated densitometry/area) of phage staining ranging from 0-255, (0 = no stain; 255 = hypothetical high) for selected regions of interest. We categorized each coronal section into different regions of interest including: cortex (CTX); corpus callosum (CC); hippocampus (HC); caudoputamen (CP); fornix (FX); corticospinal tract (CST); striatum (STR); thalamus (TH); amygdala/olfactory/cortical subplate (AMY). Not all sections contained all regions, such that 6-8 regions were de ned for each section. Due to the high temperature antigen retrieval step and the size of the multi-brain slices, adherence to the slides caused some separations and wrinkles in the tissue. Care was taken when analyzing tissue to avoid anomalous regions where tissue had tears or wrinkles, or areas where tissue was overlapped (Fig. 1B). Analyses with the C6T scFv also included percentage of region covered by clustered punctate staining. Hematoxylin nuclear staining was removed in Image J using the LAB Color Threshold option, then a black and white threshold was used to eliminate the diffuse gradient staining. The punctate C6T DAB staining was selected and the percentage of stained area (ImageJ: Area Fraction) within the selected regions of interest was calculated.

Statistical Analyses
Statistical analyses were performed in SPSS 25 (Chicago, IL, USA) and Graphpad Prism 9.5.1 (San Diego, California USA) software. Values for each antibody staining image were analyzed either as raw values or normalized to the average sham values. Uninjured shams from each time point were pooled into a single control group. In order to determine the signi cant intergroup difference, one-way analysis of variance (ANOVA) was used followed by a multiple comparison analysis using a Dunnett post hoc test. Statistical signi cance was classi ed as p < 0.05.
Correlations between behavioral scores obtained at a speci c DPI and scFv staining results obtained from brain tissue harvested at the same DPI were performed. Behavioral scores obtained at 2, 5, and 7 DPI were correlated with scFv staining levels from brain tissue harvested at 7 DPI. The r 2 values were generated by two tail Pearson bivariate correlation in SPSS.

Results
Diffuse TBI resulted in acute sensorimotor de cits and delayed onset of depressive-like behavior Mice were subjected to either one or two TBIs to model single (1 TBI) or repetitive (2 TBIs) brain injury, or to a control sham injury (sham). Diffuse TBI resulted in a signi cant injury and time effect on rotarod performance ( Fig. 2A). Mice in the 1 TBI and 2 TBIs groups had a shorter latency on the rotarod at 2, 5, and 7 DPI compared to uninjured shams (F 2, 49 = 31.01, p < 0.0001). All mice showed improvement on the rotarod as a function of time post-injury (F 2, 98 = 19.59, p < 0.0001). Mice in the 1 TBI and 2 TBIs groups had higher scores on the neurological severity score (NSS) task compared to shams (Fig. 2B). Braininjured mice had a higher NSS score at 2, 5, and 7 DPI compared to uninjured shams ( Diffuse TBI resulted in the accumulation of toxic protein variants in brain tissue Representative images of D11C scFv staining of coronal brain sections from craniectomy only sham mice and mice subjected to TBI are shown at 7, 14, or 28 DPI (Fig. 3). Dense staining of all protein variants was observed around the site of the craniectomy and brain injury at 7 DPI in both sham and TBI mice (Fig. 3A, B) re ecting high local generation and accumulation of the protein variants, though this localized staining resolved by 14 and 28 DPI (Fig. 3C-F).
A composite plot of all seven scFvs in each of the 10 different brain regions for each treatment cohort indicates the overall scFv staining intensities (Fig. 4). Results re ected the same time dependent distribution where the composite staining intensity at 7 DPI was substantially higher in regions with a closer proximity to the site of injury (Fig. 4A-C), and lower in brain regions more distant from the site of injury (Fig. 4D-J). One-way ANOVA analyzing the composite staining intensity showed signi cant differences in the cortex (CTX;  Fig. 4G), or amygdala/olfactory/cortical subplate (AMY; F 6,251 = 1.73, p = 0.12; Fig. 4J). Dunnett post hoc analysis indicated signi cant differences in the 1 TBI group and the 2 TBIs group at 7 DPI in both the CTX (Fig. 4A) and CC (Fig. 4B) compared to sham. The 1 TBI group had higher staining intensity at 7 DPI in the HC compared to sham (Fig. 4C). The 2 TBIs group had higher staining intensity in the HYP at 28 DPI compared to the sham group (Fig. 4I).
The individual protein variant staining intensity levels in different brain regions, quanti ed as a function of time post-injury, showed similar trends to those observed in the composite staining intensities presented in Fig. 4 (for full results see Supplemental Tables 1-8 and Supplemental Figs. 1-4). The C6T scFv, which binds an AD brain derived oligomeric Aβ variant, had two distinct staining patterns, a diffuse staining pattern similar to that observed with the other scFvs (Fig. 5A), and a punctate staining pattern that was observed to varying degrees in all the mice (Fig. 5B). These two staining patterns were analyzed separately (Supplemental Tables 2, 3 and Supplemental Fig. 1B, C). Signi cant differences in the CTX, the region of the brain directly below the site of TBI, were observed for all protein variants except the TDP-43 protein variant between the 7 DPI and sham groups (Supplemental Figs. 1-4). Several other signi cant differences in the 7 DPI group were also observed with selected protein variants in other brain regions near the site of injury including the CC and HC (Supplemental Figs. 1A, 2B and 3A). Only one signi cant difference in protein variant staining in a brain region further from the site of injury was observed, with the D11C tau variant between the 28 DPI and sham groups (Supplemental Fig. 2B).
Protein variant levels correlated to TBI-induced behavioral de cits.
Both protein variant levels and behavioral data showed considerable variation from mouse to mouse in all the cohorts studied including the sham group which received a craniectomy but no TBI. We correlated levels of the seven different protein variants in each of the 10 brain regions of interest in individual mice from all the groups with behavior scores measured on the same DPI. We separately correlated the C6T diffuse and punctate staining levels. There were 21 statistically signi cant correlations (p < 0.05) between protein variants in speci c brain regions and speci c behavioral de cits: 13 at 7 DPI, one at 14 DPI, and seven at 28 DPI (Table 2). Of the 13 correlations with acute behavioral de cits measured at 7 DPI, eight were with the A4 Aβ variant, two with the D11C tau variant, two with the AD-TDP3 TDP variant, and one with the D5 α-syn variant ( Table 2). All of the 13 correlations observed at 7 DPI resolved by 14 DPI, where the only signi cant correlation was with the A4 Aβ variant ( Table 2).
There were seven signi cant correlations with behavioral de cits at 28 DPI, all between either the F9T tau variant or the C6T Aβ variant ( Table 2). Levels of the F9T tau variant in three different brain regions (CC, HYP, and CST) all correlated with decreased performance in the forced swim task (FST) at 28 DPI. Levels of the C6T Aβ variant (both punctate and diffuse) in the HYP correlated with decreased performance in the FST at 28 DPI, and levels of C6T punctate staining in the CC and C6T diffuse staining in the CTX correlated with decreased performance in the novel object recognition (NOR) test at 28 DPI. There were signi cant correlations between behavioral de cits and the staining intensity of the F9T tau variant (FST- Figure 6) and the C6T Aβ variant (FST and NOR- Figure 7) at 28 DPI. While none of the 7 or 14 DPI correlations between behavior and staining intensity were statistically signi cant for either the F9T tau or C6T Aβ variants, all the correlations shifted with time and reached statistical signi cance by 28 DPI. Speci cally, F9T tau levels in the CC (Fig. 6A), HYP (Fig. 6B) and CST (Fig. 6C)

Discussion
TBI induces pathophysiological processes in the brain which can generate toxic protein variants associated with neurodegenerative diseases and corresponding behavioral de cits. Neuronal axons are particularly vulnerable to the high sheer stresses caused by TBI which can disrupt protein axonal transport mechanisms including transport of tau [4,25,26,36,47]. Stress from TBI can disrupt cellular proteostasis leading to intracellular generation and accumulation of misprocessed and sometimes toxic protein variants including variants of Aβ, tau, TDP-43 and α-syn. In humans, a TBI is a heterogenous injury that causes a broad range of morbidities including sensorimotor, affective, and cognitive de cits [33]. Clinical data suggest that some injury-induced morbidities may increase with injury severity, however, most TBI survivors, with similar injury severities, experience varied symptoms and impairments. We observed similar individual differences in our mouse cohorts. We controlled for variables such as genetic background, sex, weight, age, and location of the craniectomy. However, the individual characteristics of each mouse at the time of craniectomy and TBI resulted in varied cellular responses in each mouse brain, resulting in different behavioral outcomes and protein variant pro les in each mouse in the study.
We show that neurodegenerative disease related variants of Aβ, tau, TDP-43, and α-syn are all generated as a pathophysiological response to TBI. Numerous statistically signi cant differences in protein variant staining levels in regions adjacent to the site of injury compared to control shams were observed at 7 DPI, particularly in the corpus callosum and cortex. However, these differences essentially resolved by 14 and 28 DPI. In contrast, only a single correlation with protein variant accumulation was observed with any of the deeper brain structures, and only at 28 DPI between the D11C tau variant and the hypothalamus. However, we observed that individual mice had elevated levels of selected protein variants in the deeper brain regions, particularly in tissue collected at 28 DPI and individual mice also showed worse behavioral performance at the longer time point. We found 21 statistically signi cant correlations between protein variants in speci c brain regions and behavioral de cits, 13 re ecting acute changes (7 DPI), and seven re ecting subacute changes (28 DPI) in the brain. The changes in protein variant levels at 7 DPI re ect the disruption of cell proteostasis during the acute phase of recovery from TBI. Markedly, the A4 Aβ variant was disproportionately represented in behavioral de cits associated with the acute phase. Part of the initial acute response to TBI is an increase in neuronal expression of APP and BACE-1, a protease which cleaves APP, leading to increased Aβ levels [3,28,38,39,46]. This acute response to brain injury generates an early transcriptional burst of Aβ by neural cells, resulting in short term Aβ accumulation. This extracellular generation of Aβ has been suggested as a defensive mechanism with at least two potential functions. First, Aβ can react with brin to quickly seal ruptured blood vessels in the blood brain barrier [20] where the microvasculature of the brain is closely associated with amyloid plaques produced after injury [20]. Second, Aβ has also been shown to promote synaptogenesis, also a critically important function after brain injury [62]. Notably, the A4 scFv was generated against a small toxic synthetically generated oligomeric Aβ variant [61] that can assemble and accumulate intra-or extracellularly in the brain. The numerous acute phase correlations between A4 Aβ levels and behavioral de cits may well re ect Aβ aggregation resulting from the burst of extracellular Aβ generated during the acute response to TBI.
In contrast to the acute injury phase, during the later injury phase, the bulk of the neurodegenerative disease associated protein variants had been cleared from the brain, except for two speci c protein variants, the C6T Aβ variant and the F9T tau variant, which accounted for all of the signi cant correlations between subacute behavioral de cits and protein variant levels. C6T punctate staining in particular showed high staining levels in individual mice in several brain regions distant from the site of injury. Notably, C6T punctate staining levels were also higher in the 2 TBIs compared to 1 TBI groups in several of the deeper brain regions (HYP, AMY and CP), though there is substantial variance between individual mice and the difference did not reach statistical signi cance. Both the C6T Aβ and F9T tau variants showed a time dependent progression in their effect on behavioral outcomes where these protein variants positively correlated with acute behavioral tests (7 DPI), with a change toward a negative correlation at 14 DPI, which reached statistically signi cant negative correlations by 28 DPI.
Potential mechanistic links between TBI and increased risk of AD.
The C6T scFv was generated against an intracellularly generated Aβ variant isolated from human AD brain tissue [31]. The punctate staining pattern of C6T Aβ in the cytoplasm of pyramidal neurons in the TBI mouse brain tissue collected at 28 DPI (Fig. 5C) re ects an ongoing disruption of proteostasis.
Notably, this same staining pattern is also observed in human post-mortem AD tissue (Fig. 5D, E). Cytoplasmic accumulation of C6T reactive Aβ in neurons in the TBI mouse brain tissue at 28 DPI re ects the speci city of the C6T scFv for an intracellularly generated AD related Aβ variant, as these same Aβ variants are also present in the cytoplasm of neurons in early and late stage AD brain tissue (Fig. 5C-E). While the extracellularly generated A4 Aβ variant can accumulate as a direct consequence of the initial acute transcriptional response to brain injury, the toxic intracellular C6T Aβ variant is produced as a result of neuronal cells still under chronic stress and disrupted proteostasis after resolution of the initial injury response. Similarly, the F9T tau variant also correlated with injury and may re ect damage to neuronal axons. The white matter tracts extending from the pyramidal neurons contain high levels of tau and accumulation of neurotoxic F9T tau variants in the conductive axons of the corpus callosum can cause disruption of neuronal function [14].
The C6T Aβ and F9T tau variants correlated with long-term behavioral de cits following TBI and we previously showed that both of these variants are also excellent biomarkers for early detection of AD [6, 56, 58]. Therefore, the signi cant correlation of long-term cognitive de cits with increased levels of C6T Aβ and F9T tau variants provide compelling evidence for a pathogenic link between unresolved protein pathology resulting from TBI and early stage AD related neurodegeneration. Both the C6T Aβ and F9T tau variants have been detected in sera samples from human TBI cases; even years after injury [55], lending additional support to the validity of the mouse TBI models for replicating human TBI pathology. Notably none of the α-syn or TDP-43 variants studied correlated signi cantly with the behavioral tests at 28 DPI. The correlations of the C6T Aβ and F9T tau variants with long-term cognitive de cits in the TBI mice, along with their presence in sera samples from human TBI cases [55], and sera samples from human post-mortem and longitudinal AD cases [6, 56, 57], suggest that there is a common pathogenesis in all of these cases.
These results provide very intriguing insights into the mechanism by which TBI can lead to an increased risk of AD. It also suggests potential biomarkers that can be utilized to identify individual TBI patients at increased risk for AD and also potential therapeutic targets. We demonstrated in a mouse model of AD that selectively targeting the C6T reactive Aβ variant provides excellent therapeutic bene t, restoring neuronal integrity to wild-type levels [23]. We anticipate that there will be additional correlations between other protein variants and other behavioral tests that more closely associate with different neurodegenerative diseases, and these correlations will be the subject of further studies.
Abbreviations AD -Alzheimer's disease    Representative coronal mice brain staining detected with D11C scFv-phage grouped by Sham (craniectomy without TBI; left) or TBI (right). Tissue was collected at (A-B)7, (C-D) 14, or (E-F) or 28 days post-injury (DPI). Phage were detected with HRP conjugated secondary/DAB (brown) and nuclei detected with hematoxylin (purple). Phage staining was detected throughout the brain, though most intense staining was observed near the site of craniectomy and TBI in tissue collected at 7 DPI. The presence of the target protein variant resolved by 14 and 28 DPI. Scale bar 500 µm.

Figure 4
Combined staining intensity values of each of seven scFv phage in 10 different brain tissue regions.
Brain tissue images from each individual mouse were converted to 8-bit black and white images and inverted so phage staining correlated to brightness (0-255). Mean grey value intensity (brightness) of each brain region of interest was measured by ImageJ analysis. Values were normalized to the average grey value intensity of the corresponding brain region in sham treated mice to generate the combined values for each of the seven scFv phage. Brain tissue was collected at 7, 14, or 28 days post-injury (DPI).
Mice were subjected to either a control sham injury, 1 TBI, or 2 TBIs. Post hoc analysis of composite staining intensity of all seven scFvs showed statistically signi cant differences in the composite staining intensity at 7 DPI in TBI versus sham mice in the cortex (1 TBI p < 0.0001, 2 TBIs p < 0.0001), the corpus callosum (1 TBI p < 0.0001, 2 TBIs p = 0.002) and hippocampus (1TBI p < 0.0001). These differences resolved by 14 and 28 DPI. The only composite protein variant staining intensity that did not resolve by 14 DPI was in hypothalamus of mice subjected to 2 TBIs (p = 0.037). We also observed unresolved elevated levels of protein variants in other deeper brain regions among individual outlier cases, but these did not reach statistical signi cance. Asterisks indicate p < 0.05 signi cance compared to corresponding sham region. Error bars are ± SD.