Assessment of the Effects of Altered Amyloid-Beta Clearance on Behavior following Repeat Closed-Head Brain Injury in Amyloid-Beta Precursor Protein Humanized Mice

Abstract

Traumatic brain injury (TBI) increases the risk for dementias including Alzheimer's disease (AD) and chronic traumatic encephalopathy. Further, both human and animal model data indicate that amyloid-beta (Aβ) peptide accumulation and its production machinery are upregulated by TBI. Considering the clear link between chronic Aβ elevation and AD as well as tau pathology, the role(s) of Aβ in TBI is of high importance. Endopeptidases, including the neprilysin (NEP)-like enzymes, are key mediators of Aβ clearance and may affect susceptibility to pathology post-TBI. Here, we use a “humanized” mouse model of Aβ production, which expresses normal human amyloid-beta precursor protein (APP) under its natural transcriptional regulation and exposed them to a more clinically relevant repeated closed-head TBI paradigm. These transgenic mice also were crossed with mice deficient for the Aβ degrading enzymes NEP or NEP2 to assess models of reduced cerebral Aβ clearance in our TBI model. Our results show that the presence of the human form of Aβ did not exacerbate motor (Rotarod) and spatial learning/memory deficits (Morris water maze) post-injuries, while potentially reduced anxiety (Open Field) was observed. NEP and NEP2 deficiency also did not exacerbate these deficits post-injuries and was associated with protection from motor (NEP and NEP2) and spatial learning/memory deficits (NEP only). These data suggest that normally regulated expression of wild-type human APP/Aβ does not contribute to deficits acutely after TBI and may be protective at this stage of injury.

Introduction

An estimated 2.87 million traumatic brain injury (TBI)–related emergency department visits, hospitalizations, and deaths occurred in the United States in 2014 with 837,000 of these health events occurring in children.¹ TBI is receiving increasing attention from the research and medical communities as the connection between TBI and forms of dementia has been identified. It has been established that TBI is a major risk factor for neurodegenerative diseases including Alzheimer's disease (AD), Parkinson's disease, amyotrophic lateral sclerosis, and chronic traumatic encephalopathy (CTE).^2–7 Also, TBI shares pathological hallmarks with AD, including the development of amyloid-beta (Aβ)-plaques and hyperphosphorylated tau.⁸ Long-term survivors of a severe TBI event were shown to have elevated amyloid plaque and neurofibrillary tangle (tau) pathology.⁹

It was found that those with severe TBI showed elevated intra-axonal amyloid-beta precursor protein (APP), beta-site APP cleaving enzyme-1 (BACE1), and Aβ.¹⁰ Immediately following severe TBI in humans, diffuse but not focal amyloid plaques have been observed, as well as elevated Aβ in brain fluids.^11,12 This has been supported by a live imaging study showing elevated Aβ up to a year post-TBI.¹³ This is supported by another live imaging study, which found chronically increased Aβ deposition in the brain associated with axonal injury after TBI.¹⁴ An association of elevated Aβ with cognitive impairment also was found for those with mild TBI.¹⁵ Another live imaging study found elevated Aβ deposition in association with TBI and post-traumatic stress disorder.¹⁶ A large body of evidence shows that Aβ accumulation exacerbates/induces tau pathology in AD.^17–25 Supporting this, a clinical study by Stein and colleagues found that Aβ deposition was associated with CTE severity and tau pathology in a large subset of patients,²⁶ and it is clear that both Aβ and tau pathology are elevated by TBI events.²⁷

In wild-type animals, it has been shown that APP, its processing enzymes BACE1, and presenilin-1 (PS1), as well as Aβ, are elevated in the days after rotational and controlled cortical impact (CCI) induced TBI,^28–30 while, single severe and repeated mild TBI increased plaque deposition in APP transgenic (APPtg) mice.^31–37 This is further supported by more recent studies in wild-type mice, which found elevated Aβ post–closed-head TBI (repeat and single).^38,39

The clearance of Aβ is also likely to be involved in the response to TBI. Neprilysin (NEP) is a potent Aβ degrading enzyme, which has been shown to be relevant to AD.^40–42 NEP-deficient mice were shown to have elevated levels of Aβ in multiple studies.^{40,41,43–45} NEP expression is reported to be upregulated even years following TBI in humans.¹⁰ Most importantly, compromised NEP expression may exacerbate chronic pathology. Polymorphisms in the NEP promoter that are predicted to reduce NEP expression are associated with increased incidence of amyloid plaque pathology following TBI in humans.⁴⁶ In addition to NEP, we have found the NEP homolog, NEP2 (55% amino acid identity), is also an important Aβ degrading enzyme.⁴⁷ We found that human and mouse NEP2 degrade the major Aβ isoforms.^48,49 In addition, we have shown that NEP2 deficiency results in elevated Aβ levels in mice^48,50 and that human cerebral NEP2 is decreased early in AD progression (i.e., in mild-cognitive impairment).⁵¹ Therefore these Aβ degrading enzymes may also play a role in TBI and the associated dementias.

We set out to examine the role of Aβ acutely after TBI in a more clinically relevant model. For this, we used “humanized” APP transgenic mice, which express wild-type human APP under its normal transcriptional regulation⁵² and crossed them with NEP and NEP2-deficient mice^40,53 to facilitate elevated Aβ through reduced degradation by these enzymes. In our present study, we exposed these mice to a model of repeat closed-head TBI⁵⁴ and found that the presence of normally regulated wild-type human APP/Aβ did not exacerbate deficits post-injury. Further, NEP and NEP2 deficiencies also did not exacerbate deficits post-injury and were associated with protection from motor and cognitive dysfunction.

Methods

Animals

Several mouse lines were generated for this study. Mice possessing the entire genomic locus of normal human APP (APPtg) were purchased from the Jackson Laboratory (stock #005301; B6.129)⁵² as well as the background control strain. These APPtg mice were crossed with NEP knockout (KO; Balb/c) and NEP2 KO (B6.129) mice.^40,53 Mouse groups were APPtg with non-transgenic (Non-tg; control); APPtg.NEP^-/- with APPtg,NEP^+/+ (control); andAPPtg.NEP2^-/- with APPtg.NEP2^+/+ (control).

Rationale for the APPtg animal model

One area of interest for us was to determine the effect(s) of the presence of human Aβ on acute to sub-acute time points post-TBI. It may be that human Aβ is more pathogenic than mouse Aβ, as previously suggested.⁵⁵ Therefore, to use a better model of human pathology, we used mice producing human Aβ. Past studies using transgenic mouse models of AD have shown that the presence of these human genes exacerbate the injury and pathological outcomes.^31–37 However, these models expressed mutant forms of human APP and PS1 (or both) that cause early onset familial AD. These models also do not express these transgenes under their normal transcriptional regulation but instead constitutively express very high levels of the transgene(s). Therefore, these models are highly aberrant and are not ideal for modeling the normal process post-TBI. To address this, our models express the entire genomic locus for normal human APP in a yeast artificial chromosome.⁵² Therefore, the non-mutated APP gene is not overexpressed and retains the transcriptional control normally associated with the gene. This would allow for appropriate regulation in response to TBI.

Rationale for the endopeptidase knockout models

One major pathway for the clearance of Aβ in the brain is through direct proteolysis by endopeptidases.^42,47 NEP has been shown to be a key mediator of Aβ clearance in the brain. The NEP homolog NEP2 also degrades Aβ and evidence suggests it also contributes to Aβ clearance.^48,49,51 Knockout of either (or both) of these enzymes causes increases in Aβ levels in the brain.^40,44,48 Evidence shows that murine NEP and NEP2 degrade human Aβ.^48,56,57 Clearance pathways for Aβ have relevance to AD and we hypothesized they also may be relevant to TBI. Therefore, we crossed mouse lines deficient in the NEP or NEP2 genes with our APPtg mice to produce “humanized APP mice” lacking one of these enzymes to further elevate Aβ levels. These mice were tested in the same repeat closed-head TBI model as the APPtg and wild-type mice.

For genotyping mice, a small tail-tip biopsy (∼2 mm) was dissociated in tail lysis buffer (0.2% SDS, 0.28 mg/mL Proteinase K, 55°C ∼4 h), precipitated in isopropanol, washed in 70% ethanol, and re-suspended in ∼50 uL of Tris-EDTA buffer. Approximately 2 uL of extracted genomic DNA was used for analysis. The human APP transgene was detected by quantitative real-time polymerase chain reaction (PCR) using the standard ΔΔCt method with Eva Green (MidSci).⁵¹ Primers used were oIMR1544 (5′-CACGTGGGCTCCAGCATT-3′) and oIMR3580 (5′- TCACCAGTCATTTCTGCCTTTG-3′) for reference as well as hAPP FW-102 (5′-GGTTGACGCCGCTGTCA-3′) and hAPP RV-102 (TGG-ATT-TTC-GTA-GCC-GTT-CTG) for transgene detection (Supplementary Fig. S1A). The NEP genotype was assessed by standard end-point PCR with primers Ex-12 (5′-GAAATCATGTCAACTGTG-3′), Neo (5′-ATCAGAAGCTTATCGAT-3′), and Ex-13 (5′-CTTGCGGAAAGCATTTC-3′) at 0.8 μM each. Two reactions were run for each mouse, one for detection of the wild-type locus (using primers Ex-12 and Ex-13) and the other for detection of the knockout (using primers Ex-12 and Neo). PCR was run using standard Taq Master Mix (NEB) and the following cycle protocol: 3 min at 94°C, [30 sec at 94°C – 30 sec at 49.3°C – 30 sec at 72°C] × 30, 10 min at 72°C (Supplementary Fig. S1B). The NEP2 genotype was assessed by standard end-point PCR with primers F (5′-TGGAACTGGAGACGCATCTGG-3′), Neo-F (5′-TCCTGTCATCTCACCTGGCTCC-3′), and R (5′-TAGCTCCATCAGGTCCATTCG-3′) at 0.5 μM each altogether using standard Taq Master Mix (NEB) and the following cycle protocol: 3 min at 94°C, [30 sec at 94°C – 30 sec at 55°C – 2 min at 72°C ] × 30, 10 min at 72°C (Supplementary Fig. S1C).

Mice either received repeat impacts or repeat sham (see below). Both female and male mice were used in experiments. All mice were treated according to Institutional Animal Care and Use Committee–approved protocols and in accordance with the Guide for Care and Use of Laboratory Animals published in 1996 by the National Academy of Sciences and consistent with internal Institutional Animal Use and Care committee protocols.

TBI

We chose to follow a previous protocol described in Mouzon and colleagues in which mice were subjected to repeat closed-head impacts.⁵⁴ This is to better model the clinical situation in which the majority of people suffer multiple concussive events and not a single severe trauma directly to the brain as is done in many open-head CCI experiments. Using a controlled cortical impact device (Leica), we delivered consecutive closed-head impacts to mice at 9 weeks of age every other day for a total of five impacts (Fig. 1). Mice were anesthetized in 2% isoflurane and secured on a foam pad placed on a stereotaxic device (Kopf) mounted with a Leica Impact One device. The head was shaved and covered with thin layer of lidocaine ointment. To help measure the location of the top of the scalp, a contact sensor ground wire was attached to the tail, which completed a weak electrical circuit that produces a tone when the impactor tip touches the scalp. Preliminary experiments on mice with an open scalp were used to determine the orientation of the impactor tip for closed-head injuries to approximate the coordinates previously used.⁵⁴ The 5-mm tip was placed with the leading edge of the tip aligned 0.5 mm caudal to the caudal edge of the opening of the eye and centered medal/lateral. The impactor was then set to impact the surface of the head at a depth of 2.2 mm with a velocity of 5 m/sec and a 300 msec dwell time. Immediately post-impact or sham, mice received a low dose of meloxicam (0.4 mg/kg) subcutaneously. Sham mice were placed on the stereotaxic frame under isoflurane, shaved, and skin treated with lidocaine (as above) and kept there under isoflurane for three minutes without impacting.

FIG. 1.

Experimental design. Each node represents a single day. TBI, a single closed-head traumatic brain injury event. Color image is available online.

Rotarod analysis

One day before the first impact or sham treatment (Fig. 1), mice received pre-training on the Rotarod test. Mice were placed on a Rota Rod (Rotamex) with settings of acceleration step 0.5 rpm, max speed of 50 rpm, rot delay of 2 sec, and a maximum duration of 180 sec. The time until the mouse fell from the rod was digitally recorded. Six training trials were performed in the morning and six training trials 4 h later in the afternoon. The average time for the last three trials was used for normalized performance values for the post–injury tests. One day and 3 days post–last impact (or sham), the mice were tested with three trials on the Rotarod with the same settings as the pre-training.

Open field

Two days after the second Rotarod analysis (Day 5 post-impacts), mice were placed into a non-transparent (4′ × 4′) open field for 30 min to acclimate them to the new environment. Mouse movements were tracked and recorded using TSE Systems Top Scan. A zone 4″ from the edge was defined as the “outside” and the interior zone as “inside” for open field analysis of mouse movements.

Morris water maze

Four days after the Open Field test (9 days post-impacts), mice were placed into a water-filled 5′ diameter pool made opaque by the addition of non-toxic white paint with a clear 10-cm wide square platform 0.5 cm below the surface of the water. Mice were allowed to swim up to 1 min in order to find the platform. If the platform was not found, the mice were placed on the platform for 30 sec to orient themselves. Four trials were conducted each day for 5 days and the time to platform (latency) recorded. On the 6th day, the platform was removed and the time spent in each quadrant was recorded for 1 min for the probe trial.⁵⁸ A video tracking system TSE Systems Top Scan was used to record the movements of the mice (Supplementary Fig. S15).

Amyloid beta enzyme-linked immunosorbent assay

Mice were randomly selected at 1, 3, and 14 days post-injury or sham and anesthetized under urethane for transcardial perfusion with cold saline solution. Brains were removed and the hippocampus was dissected out for snap freezing. This tissue was later homogenized in 4 × volume/weight 5M guanidine-HCl (50 mM Tris HCl, pH 8) overnight, then diluted 10-fold in 50 mM Tris HCl (pH 8) with protease inhibitors (cOmplete-Mini, Roche) and 0.03% Tween 20, centrifuged at 16,000 × g for 30 min and supernatants collected for total Aβ extraction. An Aβ₄₀-specific enzyme-linked immunosorbent assay (ELISA) kit (Wako Chemicals, cat#: 294-64701) detecting both mouse and human Aβ as previously reported^48,50,58 was used for quantification. Data are presented as pmoles of Aβ per mg of total protein.

Statistical analysis

Statistics were analyzed using linear mixed effects models in SPSS 26.0. Specifically, in order to account for non-independence of observations (i.e., repeated measures per subject), random intercept models were fit using restricted maximum likelihood estimation for continuous variables or specifying the repeated measures factor for categorical levels. These parameterizations lead to results comparable to traditional analysis of variance (ANOVA) or t-tests, but allow for modeling heterogeneity of variance and lack of sphericity, if necessary.

Results

Human Aβ does not exacerbate post-impact deficits in the Rotarod test

It was previously shown that multiple TBI produced deficits in motor function including on the Rotarod test.⁵⁴ Therefore, we set out to assess motor function in the Rotarod test 1 day and 3 days after the last impact (or sham treatment). The data are presented as the total time on the rod or as normalized values relative to the last three pre-training trials (see the Methods section). We found a clear effect of impact on performance in the Rotarod test (Fig. 2). This is apparent when assessing total time and relative performance at both 1 day and 3 days post-impacts. It is also apparent that the presence of the human APP gene (and so the expression of human Aβ) did not exacerbate these deficits. This suggests that human Aβ expressed at near normal levels does not exacerbate motor deficits acutely post-injury.

FIG. 2.

Comparison of the effects of multiple closed-head traumatic brain injury (TBI) on performance in the Rotarod test in humanized amyloid-beta precursor protein transgenic (APPtg) and control non-transgenic (Non-tg) mice. APPtg and Non-tg mice were subjected to repeat closed-head TBI or sham treated and then tested on the Rotarod 1 (A, B) and 3 (C, D) days post-impacts. The data are presented as the mean time (A, C) or as the mean performance relative to their pre-training (B, D). Error bars represent standard error of the mean. *, **p < 0.05, 0.01 (Non-tg sham to Non-tg-TBI); #, ##p < 0.05, 0.01 (APPtg sham to APPtg-TBI).

The presence of human Aβ may affect anxiety-like behavior post-impacts

Altered anxiety is another effect reported post-TBI.⁵⁹ We assessed anxiety-like behavior in the Open Field test by measuring how much time the animal spent in the central region of an open field as opposed to the edges (see the Methods section). Mice were placed into the open field and allowed to explore for 30 min with their movements tracked by a digital tracking system. Our data suggest that the repeat TBI paradigm did not induce anxiety-like behavior, as similar times were spent in the center region of the open field for both impacted and sham groups (Fig 3). Interestingly, impacted mice expressing human APP spent more time in the center compared with the impacted wild-type mice, suggesting an effect of human Aβ on anxiety-like behavior.

FIG. 3.

Comparison of the effects of repeat closed-head traumatic brain injury (TBI) on anxiety-like behavior in the open field in humanized amyloid-beta precursor protein transgenic (APPtg) and control non-transgenic (Non-tg) mice. APPtg and Non-tg mice were acclimated to an open field with the time spent in the central region recorded. TBI, impacted; sham, sham treated. The data are presented as the mean percent time spent in the center. Error bars represent standard error of the mean. *p < 0.05.

Human Aβ does not exacerbate deficits in the Morris water maze tests

The Morris water maze is a standard test for spatial learning and memory, which is a hippocampal-dependent task that is affected in both TBI and AD models.^54,60 Mice were trained to find the location of the submerged platform for 5 days (acquisition phase), and on the 6th day the platform was removed for the probe trial. We observed an effect of the impacts on the acquisition of the platform location with impacted mice performing more poorly (Fig. 4A). The APPtg mice with TBI did not perform significantly different from their wild-type controls with TBI. The probe trial produced similar data as the time in the platform quadrant was significantly lower for impacted mice but no differences were observed between transgenic and control animals (Fig. 4B). These data suggest (as with the Rotarod analysis) the presence of human Aβ does not exacerbate deficits post-injuries.

FIG. 4.

Comparison of the effects of repeat closed-head traumatic brain injury (TBI) on spatial learning and memory in the MWM in amyloid-beta precursor protein transgenic (APPtg) and control non-transgenic (Non-tg) mice. APPtg and Non-tg mice were tested for spatial learning and memory in the MWM. TBI, impacted; sham, sham treated. (A) Acquisition of the platform location over 5 days of training. (B) Sixty-second probe trial on the 6th MWM day. Error bars represent standard error of the mean. &&p < 0.016 (APPtg-sham to all); ^^p < 0.01 (APPtg-sham to APPtg-TBI and Non-tg-TBI); @@p < 0.01 (APPtg-sham to Non-tg-TBI); *, **p < 0.05, 0.01 (Non-tg sham to Non-tg-TBI); ##p < 0.01 (APPtg sham to APPtg-TBI). Color image is available online.

Levels of Aβ₄₀ are elevated at 1 day post–repetitive closed-head TBI

It previously has been reported that TBI increases Aβ levels in animal models. As discussed above, this work has primarily been done in transgenic models of AD that are more aggressive as opposed to a more clinically relevant model like ours. Evidence also suggests that Aβ is elevated in humans suffering from TBI.^{11–13,15,16} Therefore, we set out to find evidence that our multiple-moderate impact paradigm using wild-type and transgenic mice that express normally regulated levels of APP/Aβ causes elevations in the levels of Aβ. At 1, 3, and 14 days post-impacts, a subset of mice were taken for measurements of total Aβ₄₀ in the cerebral cortex and hippocampus using a specific ELISA (see the Methods section). We found that total Aβ₄₀ levels were elevated at 1 day post-injuries compared with the Day 1 sham group and the subsequent days for the TBI group in the cerebral cortex (Fig. 5A). Also, in the hippocampus, we found elevated Aβ₄₀ at 1 day post-injury compared with the Day 1 sham group and the Day 14 post-TBI group (Fig. 5B). At Days 3 and 14 post-TBI, Aβ₄₀ levels returned to sham levels. These data support the literature in showing elevated Aβ post-TBI.

FIG. 5.

The effects of repeat closed-head traumatic brain injury (TBI) on amyloid beta levels in amyloid-beta precursor protein transgenic (APPtg) and control non-transgenic (Non-tg) mice. APPtg and Non-tg mice were sacrificed and the cerebral cortex or hippocampus was extracted and tested for Aβ₄₀ levels by specific enzyme-linked immunosorbent assay. TBI, impacted; sham, sham treated. Data are presented as mean levels of Aβ₄₀ on Days 1, 3 and 14 post-impact in the cerebral cortex (A) or hippocampus (B). Error bars represent standard error of the mean. *,**p < 0.05, 0.01.

Endopeptidase deficiencies mitigate motor deficits post–brain injuries

One day post-TBI, the Rotarod analysis showed no significant differences between NEP- or NEP2-deficient mice and NEP or NEP2 wild-type controls at Day 1 (Fig. 6A). However, at Day 3 it was found that both NEP- and NEP2-deficient mice spent more time on the Rotarod (Fig. 6C). In the NEP2 group, normalized values at Day 1 showed that impacted wild-type mice performed more poorly compared with wild-type sham mice, indicating a susceptibility to injury in the wild-type animals. However, impacted NEP2-deficient mice performed the same as non-impacted NEP2-deficient mice, suggesting a decrease in the effect of injury. The same trend is apparent in the NEP group at Day 1 and for NEP and NEP2 groups at Day 3 (Fig. 6B, 6D). These data support the findings from Figure 2 suggesting that elevated levels of Aβ do not promote pathology post-TBI and even suggest a possible protective role for the presence of this peptide. However, it should be noted that this interpretation is complicated by the fact that these endopeptidases also degrade a host of other peptides (see the Discussion section).

FIG. 6.

Comparison of the effects of repeat closed-head traumatic brain injury (TBI) on performance in the Rotarod in neprilysin (NEP), NEP2 knockout, and control mice. Amyloid-beta precursor protein transgenic (APPtg).NEP^-/- (NEPko) and APPtg.NEP2^-/- (NEP2ko) as well as control APPtg.NEP^+/+ (NEPwt) and APPtg.NEP2^+/+ (NEP2wt) mice were tested on the Rotarod 1 (A, B) and 3 (C, D) days post-impacts. TBI, impacted; sham, sham treated. The data are presented as the mean time on the rod (A, C) or as the mean performance relative to their last three pre-training trials (B, D). Error bars represent standard error of the mean. *,**p < 0.05, 0.01.

Endopeptidase deficiency does not exacerbate anxiety or deficits in spatial learning/memory and is protective in some measures

Assessments in the Open Field test showed no clear effect of endopeptidase deficiency on anxiety-related behavior (Fig. 7). However, the assessment of spatial learning and memory did indicate some effects related to the lack of NEP (Fig. 8). During the acquisition phase, NEP deficiency produced changes in performance, while NEP2 deficiency had no effect on this process (Fig. 8A, 8B). Among impacted mice, there was an overall trend for NEP-deficient mice to perform better than NEP wild-type controls which was significant on Day 4. The NEP wild-type sham group was statistically better than both impact groups (NEP knockout, NEP wild-type) on Days 2, 3, and 4. These data support the previous Morris water maze (MWM) findings (Fig. 4) that our impact protocol causes deficits in spatial learning and suggests that lack of NEP is protective. During the probe trials, the NEP-deficient impacted mice performed better than NEP wild-type impacted mice, indicating improved retention of spatial memory after injury (Fig. 8C). Alternatively, the NEP2-deficient mice did not show differences between NEP2 knockout and NEP2 wild-type genotypes. This is interesting considering there were effects of NEP2 deficiency on motor function (Fig. 6).

FIG. 7.

Comparison of the effects of repeat closed-head traumatic brain injury (TBI) on anxiety-like behavior in the open field in neprilysin (NEP), NEP2 knockout, and control mice. Amyloid-beta precursor protein transgenic (APPtg).NEP^-/- (NEPko) and APPtg.NEP2^-/- (NEP2ko) as well as control APPtg.NEP^+/+ (NEPwt) and APPtg.NEP2^+/+ (NEP2wt) mice were tested in an open field with the time spent in the central region recorded. TBI, impacted; sham, sham treated. The data are presented as the mean percent time spent in the center. Error bars represent standard error of the mean. *p < 0.05.

FIG. 8.

Comparison of the effects of repeat closed-head traumatic brain injury (TBI) on spatial learning and memory in the Morris water maze (MWM) in neprilysin (NEP), NEP2 knockout, and control mice. Amyloid-beta precursor protein transgenic (APPtg).NEP^-/- (NEPko) and APPtg.NEP2^-/- (NEP2ko) as well as control APPtg.NEP^+/+ (NEPwt) and APPtg.NEP2^+/+ (NEP2wt) mice were tested for spatial learning and memory in the Morris water maze (MWM). TBI, impacted; sham, sham treated. (A, B) Acquisition of the platform location over 5 days of training for NEP altered (A) and NEP2 altered (B) mice. Data are represented as means ± standard error of the mean (SEM). ††p < 0.01 (either sham to either TBI); ‡, ‡‡p < 0.05, 0.01 (NEPwt-sham to either TBI groups); ##p < 0.01 (NEPwt-TBI to NEPko-TBI); @, @@p < 0.05, 0.01 (NEPko-sham to NEPwt-TBI); **p < 0.01 (NEPwt-sham to NEPwt-TBI). (C) Sixty-second probe trial on the 6th MWM day. Data are represented as means ± SEM. *, **p < 0.05, 0.01. Group sizes are the same for both acquisition and probe data. Color image is available online.

NEP deficiency produced elevated levels of Aβ₄₀ in the cerebral cortex and hippocampus

As discussed, a major target for degradation by NEP and NEP2 is the Aβ peptide. Therefore, we measured the levels of Aβ₄₀ in the brains of our mice. We assessed levels of this peptide in both cerebral cortex and hippocampus at Day 14. Aβ₄₀ levels were significantly increased in NEP-deficient mice compared with NEP wild-type mice in both the TBI and sham groups (Fig. 9A, 9B). As predicted by our analysis of Aβ₄₀ in APPtg and Non-tg mice (Fig. 5), we did not see an elevation of Aβ₄₀ in the TBI group compared with sham at Day 14. For NEP2, we did not see a significant increase in Aβ₄₀ levels in NEP2-deficient mice in either the cerebral cortex or hippocampus, nor did we see an effect of TBI (Fig. 9A, 9B). Because of the Aβ isoform assayed and the brain regions assessed these results regarding NEP2 deficiency are not entirely surprising (see the Discussion section).

FIG. 9.

The effects of repeat closed-head traumatic brain injury (TBI) on amyloid beta levels in amyloid-beta precursor protein transgenic (APPtg) mice with altered neprilysin (NEP) or NEP2 levels. APPtg mice that were deficient for NEP or NEP2, knockout (ko), as well as NEP or NEP2 wild-type (wt) controls were sacrificed at Day 14 post-impact and the cerebral cortex or hippocampus was extracted for assay of Aβ₄₀ levels by specific enzyme-linked immunosorbent assay. TBI, impacted; sham, sham treated. Data are presented as mean levels of Aβ₄₀ in the cerebral cortex (A) or hippocampus (B). Error bars represent standard error of the mean. **p < 0.01.

Analysis of sex as a variable

We have included all our data with females and males in separate groups (see Supplementary Figs. S2–S12) for side-by-side comparisons. We also estimated models with sex, as well as interactions of sex with other variables, as variables in our analyses. With the exception of the Open Field test, we did not find a significant interaction of sex with treatment and genotype. In the open field, we found that NEPwt female mice with TBI spent more time in the center of the open field compared with males of the same group (Supplementary Fig. S9). We did find some sex differences related to Rotarod performance within the relative performance data. APPtg male mice performed significantly better than APPtg female mice as a whole (combining TBI and sham groups). APPtg male mice also performed better than male Non-tg mice (Supplementary Fig. S13A). While these differences were significant, the change in performance was marginal (12%). Further, within the Rotarod relative performance data for the NEP and NEP2 gene altered groups, we found that male mice deficient for NEP (NEPko), as a whole, performed better (by 20%) than female NEPko mice. Within the NEP2 altered mouse group, male mice that were wild-type (NEP2wt) also performed better than females of the same group (by 25%; Supplementary Fig. S13B).

Discussion

Our study indicates that the mere presence of human Aβ is not harmful acutely after repeat closed-head TBI even though many believe the human form is more amyloidogenic. Also, the disruption of Aβ clearance by endopeptidase deficiency (NEP and NEP2 knockouts) did not exacerbate deficits post-injury and was found to be protective with regard to motor function, spatial learning/memory, and possibly anxiety. These findings are at odds with a large body of evidence that Aβ is neurologically harmful.

It should be noted that, while the pathologic role of Aβ in the chronic development of dementia is well established,⁴⁷ the acute effects of Aβ are less well understood. As stated above, the conventional wisdom is that Aβ contributes to pathology acutely after TBI. However, the TBI field is somewhat conflicted on the issue of acute Aβ toxicity.⁶¹ In line with a deleterious role for Aβ, Loane and colleagues showed that BACE1 KO and PS1 inhibition (both lowering Aβ) protect against pathology after CCI.²⁸ However, this is contradicted in a study by Mannix and colleagues that showed BACE1 KO mice were more vulnerable to this type of TBI.³⁰ This study was followed up by a second study showing that the injection of Aβ₄₀ reduced some of the deficits after TBI in BACE1 KO mice, further supporting a protective role for Aβ.⁶² Additionally, Corrigan and colleagues showed that knockout of APP exacerbated pathology following “diffuse” CCI,⁶³ again supporting a possible beneficial acute effect of Aβ. However, a follow-up study indicated that the lack of soluble αAPP, not Aβ, was the primary driver of exacerbated pathology.⁶⁴ A study on spinal cord injury in mice found that reduced Aβ also exacerbated acute pathology.⁶⁵ Finally, and most importantly, a human clinical study of severe TBI found that Aβ levels positively correlated with improved neurological status.⁶⁶

Our data would support a potential protective role for Aβ acutely after repeat closed-head TBI and so may be part of the normal adaptive response to injury. This is consistent with our finding of elevated Aβ₄₀ in NEP-deficient mice (Fig. 9). However, we did not observe elevated Aβ₄₀ in NEP2-deficient mice. Previously, we have shown elevated Aβ in NEP2ko mice⁴⁸; however, it should be noted that in non-transgenic mice, we did not observe an effect of NEP2 deficiency on Aβ levels in the cerebral cortex. The effects on Aβ levels were most notable in the hippocampus and brainstem/diencephalon. Further, the effects of NEP2 deficiency on Aβ levels were previously found to be more prominent on the Aβ₄₂ isoform as opposed to Aβ₄₀, which we measured in the current study. Therefore, these data are not entirely inconsistent with our previous findings.

The interpretation of these data using NEP- and NEP2-deficient mice is complicated by the existence of other substrates for these enzymes besides Aβ.⁴² NEP is a cell surface associated metalloprotease that degrades signaling peptides in the periphery and central nervous system. Substrates include (but are not limited to) enkephalin, bradykinin, substance-P, and angiotensin.⁶⁷ NEP substrates are redundantly degraded by other proteases (including NEP2, which has similar targets, particularly in rodents).^68,69 NEPko mice appear overtly normal; however, these mice were reported to show increased vulnerability to peripheral endotoxin⁷⁰ and it is thought that NEP acts to control inflammation through the degradation of immunoreactive peptides like substance-P.⁷¹ NEP2ko mice also are overtly normal but males are reported to be less fertile.⁵³ We must be careful in our data interpretations as NEP/NEP2 may also affect inflammatory processes related to TBI. It is noteworthy, however, that the examples of the anti-inflammatory properties of NEP or NEP2 would lead to a prediction of increased vulnerability to TBI in knockout mice, which we have not seen. It is interesting that NEP deficiency affected both motor and spatial learning/memory post-injury while NEP2 deficiency only affected motor performance (Fig. 6 and Fig. 8). To our knowledge this is the first study to examine to role of NEP and NEP2 in a TBI animal model. It should also be noted that another class of metalloproteases (matrix metalloproteases) also have been implicated in the response to brain injury.⁷²

Regarding the Rotarod performance, the clear difference between APPtg TBI and sham that was seen in APPtg and Non-tg mice (Fig. 2) is no longer significant in the genome control groups for NEP/NEP2 altered mice (i.e., NEP and NEP2 wild-type; Fig. 6). In most cases, a trend is observed in which the TBI group performed worse than the sham group. We hypothesize larger groups sizes would have resulted in statistically significant differences. However, there is not even a clear trend towards reduced performance on the Rotarod after TBI on Day 1 in the NEP wild-type mouse groups (Fig. 6A). We hypothesize that this may be attributable to strain genetic background differences between the mice used in Figure 2 and Figure 6. We speculate that there may be more resilience in the Balb/c background contribution from the NEP knockout mice. Also, this observation seems to be primarily driven by the females in the group (Supplementary Fig. S7A, S7C).

The use of both males and females can add to variability due to estrous cycle effects as well as other gender differences. Therefore, we conducted an analysis of sex as a variable in our data. Our assessment found an interaction of sex with the Open Field test (Supplementary Fig. S9) and in Rotarod performance (Supplementary Fig. S13). It is unclear as to why the female versus male performance ratios would differ between the NEP and NEP2 gene altered groups (i.e., males performing better in NEPko mice and in NEP2wt mice) in the Rotarod test. Mouse strain background differences could contribute to this. Also, NEP expression has been linked to androgen expression.^73,74 Our analysis did not find that these sex differences contributed to observed significant differences between combined groups in our Rotarod tests (Fig. 2 and Fig. 6).

One limitation of our study is that it does not provide information on classical markers of AD pathology like amyloid plaques or oligomer formation after TBI. This is because our model of normal human APP expression is not reported to produce plaques due to a relatively low level of Aβ production. For example, our previous studies have found approximately 100-fold elevations in levels of Aβ in a more traditional type of APPtg model (TASD41) compared with the APPtg mice used in this study.⁴⁸ We also have not yet assessed axonal APP and other hallmarks of TBI pathology in these mice. These are areas of future interest. Lastly, it should be mentioned that our study used relatively young (9 weeks) mice for our experiments. Therefore, our work is most relevant to young adults and does not directly test the role of human Aβ, NEP, or NEP2 in aged individuals.

Finally, our data replicate the findings by Mouzon and colleagues, as we also found deficits in motor and cognitive function as a result of repeat closed-head impacts using a controlled cortical impact device with a similar impact schedule and location.⁵⁴ While our study also used Rotarod analysis we did not use the same measure of spatial learning. Our protocol used the Morris water maze as opposed to their use of the Barnes maze; however, we achieved similar results. We also additionally probed the affective state of the animal in the Open Field test (Fig. 3 and Fig. 7) and to some extent explored the effect of repeat closed-head TBI on a measure of episodic-like memory (Novel Object Recognition Test or NORT; see Supplementary NORT Data) in which we interestingly found no effect of injuries on NORT (Supplementary Fig. S14). We had some difficulty with the NORT analysis for this study and had to make adjustment to the interval between acquisition of the old objects and exposure to the new object as well as the types of objects used. Therefore, we only generated relevant data for the later cohorts done for the comparisons of wild-type mice with APPtg mice which came after our NEP and NEP2 experiments were completed.

Contrary to what was expected regarding anxiety, open field analysis of the wild-type versus APPtg mice in the impact group indicates that anxiety may be decreased in APPtg mice after impact (Fig. 3). This is somewhat similar to findings by Petraglia and colleagues that showed increased exploratory behavior in the elevated plus maze at later time points post-injuries,⁵⁹ which is comparable to more time spent in the center region of the open field (i.e., lower anxiety). However, it should be noted that the apparent reduction in anxiety after TBI (Fig. 3) should be interpreted with caution because the time in the center trends upwards for the APPtg mice in general compared with Non-tg. Thus, this may represent a general effect on anxiety due to the APP transgene rather than an effect of the transgene on the response to TBI. It is notable that the decrease in anxiety observed in APPtg versus Non-tg mice (with TBI) was primarily driven by the males in the group (Supplementary Fig. S4). Finally, no effect of TBI on anxiety in the open field was observed in the NEP/NEP2 cohort (Fig. 7). We also note that NEP2ko produced reduced anxiety but only in the sham group.

In conclusion, using a humanized model of Aβ without any aberrant expression/mutations we found that human Aβ does not exacerbate deficits acutely following TBI and suggest that Aβ may have some acute protective properties. In addition, we have replicated the findings of Mouzon and colleagues using their repeat closed-head impact paradigm for modeling TBI. Future experiments will focus on pathological assessments in these mice post-injuries as well as the effect of elevated NEP levels through transgenesis.⁷⁵

Funding Information

We would like to acknowledge funding from National Institutes of Health/National Institute of Neurological Disorders and Stroke R21 NS093570 (RAM, DAK).

Author Disclosure Statement

No competing financial interests exist.

Supplementary Material

Supplementary NORT Data

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Figure S4

Supplementary Figure S5

Supplementary Figure S6

Supplementary Figure S7

Supplementary Figure S8

Supplementary Figure S9

Supplementary Figure S10

Supplementary Figure S11

Supplementary Figure S12

Supplementary Figure S13

Supplementary Figure S14

Supplementary Figure S15