Abrogation of atypical neurogenesis and vascular-derived EphA4 prevents repeated mild TBI-induced learning and memory impairments

Brain injury resulting from repeated mild traumatic insult is associated with cognitive dysfunction and other chronic co-morbidities. The current study tested the effects of aberrant neurogenesis in a mouse model of repeated mild traumatic brain injury (rmTBI). Using Barnes Maze analysis, we found a significant reduction in spatial learning and memory at 24 days post-rmTBI compared to repeated sham (rSham) injury. Cell fate analysis showed a greater number of BrdU-labeled cells which co-expressed Prox-1 in the DG of rmTBI-injured mice which coincided with enhanced cFos expression for neuronal activity. We then selectively ablated dividing neural progenitor cells using a 7-day continuous infusion of Ara-C prior to rSham or rmTBI. This resulted in attenuation of cFos and BrdU-labeled cell changes and prevented associated learning and memory deficits. We further showed this phenotype was ameliorated in EphA4f./f/Tie2-Cre knockout compared to EphA4f./f wild type mice, which coincided with altered mRNA transcript levels of MCP-1, Cx43 and TGFβ. These findings demonstrate that cognitive decline is associated with an increased presence of immature neurons and gene expression changes in the DG following rmTBI. Our data also suggests that vascular EphA4-mediated neurogenic remodeling adversely affects learning and memory behavior in response to repeated insult.

www.nature.com/scientificreports/ or drug-mediated suppression of this response has been shown to impair spontaneous cognitive recovery after moderate lateral fluid percussion injury (FPI) 15 . However, additional studies suggest that FPI-induced neurogenesis may adversely affect long-term outcomes and contribute to the development of epileptogenesis 17 . Conflicting evidence also exists regarding the influence of neurogenic ablation on pilocarpine-induced seizures and epilepsy-associated cognitive deficits 18 . Murine models have also established that ischemia-induced aberrant neurogenesis in the hippocampus is associated with cognitive decline 19,20 which is ameliorated following either pharmacological or genetic neurogenic ablation strategies 21 .
Atypical proliferation and migration of immature neurons to the outer granule layer of the dentate gyrus and hilus occurs in several models of single moderate TBI 22,23 , however, it remains unclear whether these TBIinduced changes are model dependent, benefit recovery from injury or mediate the establishment of an aberrant state in the hippocampal circuitry. Compared to single mild TBI, new findings suggest that repeated mild TBI (rmTBI) in a murine model results in significant chronic learning and memory deficits [24][25][26][27][28][29] . In the present study, we evaluated whether these effects correlate with substantial neurogenic changes in the DG using BrdU fate labeling, the causal relationship between these events and the potential mechanisms involved using cell-specific knockout mice. Strong evidence supports the role of vascular-derived ephrin type-A receptor 4 (EphA4) in the regulation of hippocampal neurogenesis 30,31 . Our recent findings implicate EphA4 as critical player in cerebrovascular function [32][33][34] . EphA and EphB class receptors, the largest group of receptor tyrosine kinases, and their membrane bound ephrin ligands mediate cell-to-cell contact signaling involved in numerous cell functions including cell migration, proliferation, survival, axon guidance, synaptogenesis, and vascular remodeling among other roles [35][36][37][38] . Using endothelial cell (EC)-specific EphA4 knockout (KO) mice (EphA4f. /f /Tie2-Cre) and wild type (WT) mice (EphA4f. /f ) 32,33 , the current study sought to advance our understanding of Eph signaling in aberrant neurogenesis as it relates to brain trauma. Our findings uncovered the direct involvement of atypical neurogenesis in the observed hippocampal-dependent learning and memory impairments following rmTBI. We further provide evidence that vascular-derived EphA4 receptor tyrosine kinase is a key player in this response.

Methods
Animals. All mice were housed in an AAALAC accredited, virus/specific antigen-free facility with a 12 h light-dark cycle; food and water ad libitum. Male CD1 mice were purchased from Charles Rivers and reared until age P60-P90 for experimentation. Female EphA4f. /f and male EphA4f. /f /Tie2-Cre +/− on the CD1 background were bred as previously described 32,33 to generate the appropriate number of male mice, reared until age P60-P90, required for each experiment. All experiments were conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were conducted performed under the approval of the Virginia Tech Institutional Animal Care and Use Committee (IACUC; #17-093) and the Virginia-Maryland Regional College of Veterinary Medicine.
Repetitive TBI and sham protocol. Adult male mice (P60-90) were anesthetized using ketamine (100 mg/kg) and xylazine (10 mg/kg) subcutaneous injection and positioned in a stereotaxic frame. Body temperature was monitored and maintained at 37 °C with a controlled heating pad set and rectal probe (Harvard Apparatus). Injury was induced by a program controlled cortical impactor (Φ = 5-mm flat tip) connected to an eCCI-6.3 device (Custom Design and Fabrication, LLC) at a velocity of 5.0 m/s, depth of 1.0 mm, and 200 ms impact duration. The flat impactor tip was positioned directly between lambda and bregma. Following injury, the incision was closed using Vetbond tissue adhesive (3 M, St. Paul, MN, USA) and the animals were placed into a heated cage and monitored until fully recovered from anesthesia. Repeated surgeries for TBI and sham were conducted once every 2 days for a total of 5 surgeries 26 .
BrdU and the Ara-C administration. Bromodeoxyuridine (BrdU; 50 mg/kg/day) was injected on day 0-5 during the repeated injury paradigm 39 . For AraC Infusions, Alzet Mini-Osmotic Pumps Model 1007D (DURECT Corporation, Cupertino, CA) were used to provide continuous systematic delivery of the vehicle saline control or 2% AraC (Sigma, St. Louis, MO, USA) after subcutaneous implantation 40 . Infusions were performed 7 days prior to rmTBI and rSham at which time the pumps were removed on the day of surgery.
Barnes maze evaluation. Mice were evaluated for learning and memory utilizing the Barnes Maze (92 cm diameter, 20 holes, 5 cm holes) (Maze Engineers, Boston, MA USA) and the Ethovision XT software (Noldus, Leesburg, VA USA) at the same time each day in the early am 26 . Visual cues were provided on the wall above the table and an overhead light was used for illumination and the maze is cleaned to remove residual olfactory cues from previous runs. Mice were placed in the center of the table under an opaque jar for 20 s until the start of the trial. Mice were first pre-trained to enter the escape box by allowing them to freely explore the maze for 5-min total with no anxiety-inducing lights. During this training phase, mice were prompted to enter the hole 3 times, once at the very beginning and once at the very end of the 5-min trial if they had not already entered the escape box. No analyses were collected during the training phase. The following three days (days 21-23 post primaryinjury), mice were evaluated for distance, time, and the number of errors to find the escape box each day for 180 s/trial in 3 trials/day for a total of 9 trials. Primary errors were calculated every time the subject poked its head into a hole that was not the target hole. During each of these acquisition trials, if the mice did not enter the escape box, they were prompted to enter and remain for 30 s before returning to their home cage. The average of the 3 daily trials were reported. tissue processing and staining. Mice were euthanized by overdose of isoflurane. Brains were harvested at both 9 and 24 days post-primary injury (dppi) and immediately fresh frozen and embedded in OCT. Serial Scientific RepoRtS | (2020) 10:15374 | https://doi.org/10.1038/s41598-020-72380-1 www.nature.com/scientificreports/ coronal sections were processed on a cryostat (Cryostar, Fisher Scientific) using five 30 μm coronal sections spaced ten sections apart and placed in the − 80 °C freezer as previously performed 32,[41][42][43]  Quantitative real-time PCR. Total RNA from the microdissected dentate gyrus was isolated according to manufactures instructions using TRIzol reagent (Ambion). Quantification of RNA was done using spectrophotometer ND-1000 (NanoDrop). We then used 1 µg of total RNA for reverse transcription reaction using iScript cDNA synthesis kit (Biorad, Hercules, CA). Next qRT-PCR analysis, used 50 ng cDNA per reaction and was amplified using iTaq Universal SYBR Green Supermix (Biorad, Hercules, CA). We calculated expression changes using ΔCq values relative to the internal control gapdh for each sample. Data was represented as relative expression, which was calculated by normalizing each value to the average WT sham for each gene. All primers ( Table 2) showed a primer efficiency range from 87 to 113%.
Statistical analysis. GraphPad Prism, version 7 (GraphPad Software, Inc., San Diego, CA) was used to graph all data. Multiple comparisons were done using repeated measures two-way ANOVA for behavior or one-way ANOVA for stereology where appropriate followed by Bonferroni post-hoc test for multiple pairwise examinations. Significant changes were identified if p was less than 0.05. Mean values are represented together with the standard error of mean (SEM). Sample size was determined based on G*Power 3 (Universitat Dusseldorf, Germany) to retrieve sample size with an acceptable power range between 80-90%. All animals were coded and a double-blinded strategy was used for statistical analysis.

Repeated mild TBI (rmTBI) induces learning and memory deficits and enhanced neurogenesis.
To investigate the contribution of neurogenesis to learning and memory deficits following rmTBI, we utilized a previously established mouse model 26 . We subjected mice to five mild closed head impacts, spaced 2 days apart followed by Barnes Maze analysis at 21-23 days post-primary injury (dppi) and histological testing using bromodeoxyuridine (BrdU) labeling at 9 and 24 dppi (Fig. 1G). We observed changes in expression of GFAP immunoreactivity in the cortex, entorhinal cortex, and hippocampus of rmTBI mice ( Fig. 1D-F) as compared to repeated sham-injured (rSham) mice ( Fig. 1A-C). We also found prominent GFAP immunoreactivity on CD31-positive vessel in the dentate gyrus (DG) at 24 dppi rmTBI ( Fig. 1D1-F1) and cortex ( Fig. 1D2-F2). In addition, we developed a new, shortened protocol to test for learning and memory deficits using the Barnes Maze (Fig. 1G). Using this method, rmTBI mice displayed a significant increase in the number of primary errors ( These findings confirm previous studies that show enhanced GFAP immunoreactivity as well as learning and memory deficits following rmTBI 26 . Next, BrdU was injected at 0-4 dppi to label and track the fate of neural stem/progenitor cells (NSPCs) in the DG quantified by using non-biased stereology. At 24 dppi post-rSham, we found a significant reduction in the number of BrdU-positive cells in the DG compared to 9 dppi rSham (Fig. 1L). The observed reduction of BrdU-labeled cells over time in rSham mice was prevented in rmTBI mice (F (1,22) = 5.133; p = 0.03). The population of BrdU-labeled cells that remained in the rmTBI DG displayed doublelabeling with Prox1 (Fig. 1M), a marker for proliferating neuroblasts and immature neurons 44 , but not NSPC marker Sox2 45 (Fig. 1N) or mature neuronal marker NeuN 46 (Fig. 1O). These data demonstrate that learning and memory impairments correlate with an atypical increase in the number of immature neurons in the DG.
Aberrant neurogenesis plays a role in learning and memory deficits following rmTBI. We evaluated whether the atypical changes in hippocampal neurogenesis contribute to cognitive dysfunction as a result of rmTBI. To test this, we pharmacologically suppressed neurogenesis by delivering cytosine arabinoside (AraC) or vehicle control 15,47 using subcutaneous implants of mini-osmotic pumps and daily i.p. injections of 50 mg/kg BrdU for 7 days prior to rmTBI or rSham ( Fig. 2A). After 7-days infusion (0 dppi), we show AraC suppressed BrdU immunolabeling in the DG compared to vehicle-infused mice (Supp. Figure 1). Using this strategy, we then performed rmTBI or rSham following removal of AraC or vehicle osmotic pumps and evaluated histological and behavior changes. Similar to our initial findings, we observed vehicle-treated rmTBI mice showed a sig- This demonstrates that early suppression of neurogenesis can prevent rmTBI-induced expansion of the Prox-1-positive population in the DG. Next, we tested whether abrogation of aberrant neurogenesis could alleviate learning and memory impairments in mice subjected to rmTBI. Using Barnes Maze, we found vehicle-treated rmTBI mice display a significant increase in the number of primary errors at 23 dppi compared to vehicle-treated rSham (Fig. 3A). Interestingly, we observed that Ara-C treated mice attenuated this effect (F (3,37) = 3.33; p = 0.02), suggesting atypical neurogenesis may play an essential role in rmTBI-induced cognitive decline. No difference was observed in latency (F (3,40) = 0.77; p = 0.51), or distance traveled (F (3,38) = 1.14; p = 0.34) between groups. We then quantified the number of cFos-immunolabeled cells in the DG, as a marker of neuronal and functional activity in the brain 48 . Vehicle-treated rmTBI mice displayed a significant increase in the number of cFos-positive cells in the DG (Fig. 3B,F) compared to vehicle-treated rSham (Fig. 3B,E) that was not present in AraC-treated mice (Fig. 3B,G,H)   While we observed a significant increase of Prox1-positive cells in the dentate hilus of vehicle-treated rmTBI mice, this effect was not influenced by AraC treatment (Fig. 3D) (F (3,26) = 17.37; p < 0.0001). Taken together, these data imply that the increased neurogenic response and hyperactivity in the DG contribute to learning and memory deficits following rmTBI.   (I-L) Quantified mRNA expression of total DG tissue using qPCR for FGF2, MCP-1, Gja1 (Cx43), and TGFβ, respectively. All genes were normalized to GAPDH then represented relative to WT rSham levels. *p < 0.05; **p < 0.01, ***p = 0.001. Scale = 100 µm in (G,H); 20 µm in inset. n = 5-10 mice per group for stereology and n = 10-16 for behavior. qPCR was performed using biological triplicates. www.nature.com/scientificreports/ tative real-time PCR (qPCR) for several genes well-known to be involved in neurogenesis [50][51][52][53] . DG tissue was microdissected at 24 dppi from rSham and rmTBI mice. While no difference was found in FGF2 expression (

Discussion
Long-term cognitive impairments can result from repeated traumatic insult to the brain 54 . Our findings demonstrate, using a murine model of rmTBI 26 , that learning and memory deficits occur as early as 24 days post-initial injury. This neurological change coincides with an increased presence of BrdU-labeled Prox1+ neuroblasts or immature neurons in the DG that is not seen at 9 dppi. The loss of BrdU-labeled cells originating from the NSPC pool overtime is a normal process of neurogenesis. However, the sustained presence of BrdU/Prox1-positive cells in the DG at 24 dppi may result from enhanced proliferation and/or survival of the neuroblast population. This is evident due to the fact that the resulting pool of fate-labeled Sox2+ NSPCs and NeuN+ mature neurons were unchanged following rmTBI suggesting the immature neuron population is selectively affected by rmTBI. Transient suppression of neurogenesis, using AraC infusion to ablate dividing NSPCs prior to rmTBI, prevented the expansion of the Prox-1+ cell population at 24 dppi. This coincided with restoration of proper hippocampaldependent learning and memory function and attenuation of enhanced cFos immunolabeling in the DG that represents neuronal hyperactivity 55 . While the expression of immediate early gene, cFos, is associated with learning and memory 48 , atypical expression can be found in areas of the brain that are hyperexcited or following seizure induction such as the DG, cortex, hippocampus and limbic system 56 . Changes in cFos have also been observed in the brain following stress or injury 57 . Here we show increased expression in the DG in response to rmTBI is positively correlated with behavioral deficits. Moreover, ectopic hilar migration of Prox1-positive cells influences DG function and has been observed following FPI model of TBI 15,58 . We also show ectopic migration of Prox1 granular neurons into the hilus following rmTBI was not prevented by AraC treatment, suggesting this occurrence may not contribute to memory impairments assessed by Barnes Maze testing. Our findings further validate the negative role of aberrant hippocampal neurogenesis in functional cognitive outcomes as a result of repeated mild trauma to the brain. Vascular involvement in hippocampal neurogenesis has been postulated to regulate neural progenitor cell behavior and responses to external stimuli [59][60][61] . Blood vessels act as physical substrates for proliferating neuroblasts during tangential migration in the DG 62 . Vascular-neuroblast interactions are a crucial niche component necessary for cell division and migration status of immature neurons [62][63][64] . Vascular endothelial cell (EC) expression of EphA4 receptor tyrosine kinase has been implicated in regulating neurogenesis in the DG 31 . Cell-to-cell contact involving membrane bound ephrin ligand and Eph receptor signaling plays a key role in neurogenesis 39,65,66 . We previously generated cell-specific EphA4 KO mice using the Tie2 promoter (EphA4f. /f / Tie2-Cre), which results in deletion of EphA4 in ECS and hematopoietic cells 32,33 . While loss of EC-specific EphA4 does not result in observable changes in neurogenesis using fate labeling in un-injured mice, we find that rmTBI-induced increase in BrdU/Prox1 double-positive cell population was attenuated in KO mice. This correlated with a rescue of learning and memory deficits and neuronal activity as quantified by cFos expression in the DG. These findings suggest that vascular expression and function of EphA4 mediates overproduction of immature neurons in the DG following rmTBI. Given the loss of EphA4 in hematopoietic cells in our KO mice, we cannot rule out the possibility of peripheral immune cell-specific contributions which will require further investigation. However, the BBB has not been shown to be disrupted in this model.
Lastly, rmTBI results in chronic downregulation of MCP-1, Gja1 (Cx43) and TGFβ mRNA expression in the microdissected DG of WT mice, which was not seen in KO mice. Importantly, rmTBI KO mice show a statistically significant attenuation in Cx43 downregulation compared to WT rmTBI suggesting Cx43 may mediate ECspecific EphA4 effects following injury. Previous studies show hippocampal multipotent progenitor cells display reduced levels of Cx43 during intermediate stage of neuronal differentiation 67 . This may reflect the change in DG composition which shows increased immature neuron production after rmTBI. Studies also show using single cell RNA sequencing of the mouse DG, that Cx43 is downregulated in proliferating neural progenitor cells 68 .  www.nature.com/scientificreports/ While the expression of EC-specific Cx43 in the DG has not been investigated, the role of vascular influence in hippocampal neurogenesis through gap junction (GJ)-associated Cx43 remains to be established. Interestingly, our unpublished findings show enhanced expression of Cx3 protein in cultured EphA4-null ECs compared to WT ECs (data not shown). Further studies are needed to determine whether EC-specific EphA4 activation may mediate expansion of the neuroblast or immature neuron population by suppressing Cx43 expression and/or function during neuroblast-vascular coupling. While the controversial role of aberrant neurogenesis in the pathogenesis of TBI remains under investigation, our findings highlight a negative role in cognitive outcomes in a murine model of repeated mild injury. Through fate labeling, we show that the immature neuron population is selectively influenced by rmTBI and that vascular expression of EphA4 may mediate this phenotype by regulating Cx43 expression in the DG. Overproduction of immature neurons correlates with hyperactivity as seen through enhanced presence of cFos DG staining. This effect directly coincided with the number of errors made by rmTBI mice on the Barnes Maze spatial learning task which was rescued following neurogenic suppression using AraC. This implies that aberrant neurogenesis may interrupt the normal hippocampal circuitry required for proper learning and memory tasks. Findings from this study uncover a novel mechanism regulating the atypical neurogenic response to rmTBI which may be exploited for future hypothesis-driven research.