Brain Trauma Induces Massive Hippocampal Neuron Death Linked to a Surge in β-Amyloid Levels in Mice Overexpressing Mutant Amyloid Precursor Protein

Abstract

Although brain trauma is a risk factor for Alzheimer’s disease, no experimental model has been generated to explore this relationship. We developed a model of brain trauma in transgenic mice that overexpress mutant human amyloid precursor protein (PDAPP) leading to the appearance of Alzheimer’s disease-like β-amyloid (Aβ) plaques beginning at 6 months of age. We induced cortical impact brain injury in the PDAPP animals and their wild-type littermates at 4 months of age, ie, before Aβ plaque formation, and evaluated changes in posttraumatic memory function, histopathology, and regional tissue levels of the Aβ peptides Aβ_1–40 and Aβ_1–42. We found that noninjured PDAPP mice had impaired memory function compared to noninjured wild-type littermates (P < 0.01) and that brain-injured PDAPP mice had more profound memory dysfunction than brain-injured wild-type littermates (P < 0.001). Although no augmentation of Aβ plaque formation was observed in brain-injured PDAPP mice, a substantial exacerbation of neuron death was found in the hippocampus (P < 0.001) in association with an acute threefold increase in Aβ_1–40 and sevenfold increase in Aβ_1–42 levels selectively in the hippocampus (P < 0.01). These data suggest a mechanistic link between brain trauma and Aβ levels and the death of neurons.

Although circumstantial evidence suggests that traumatic brain injury is a risk factor for Alzheimer’s disease (AD), the mechanisms underlying this relationship remain unknown. Previously, postmortem histopathological analysis of brains from boxers with dementia pugilistica (punch-drunk syndrome) revealed neurofibrillary tangles and diffuse plaques composed of β-amyloid peptides (Aβs) similar to AD lesions. 1,2 Indeed, a single incident of brain trauma may result in a widespread deposition of Aβ, 3,4 and a history of brain trauma increases an individual’s risk for AD. 5 Moreover, marked accumulations of β-amyloid precursor proteins have been seen after brain trauma in humans and experimental animals. 6,7,8,9 Although this suggests that ample substrates are available for pathological Aβ production after trauma, the investigation of mechanisms whereby brain trauma induces deposition of Aβ has been hampered by the inability of experimental models of brain trauma in rodents to produce Aβ-containing plaques. 6,7 Although there are several potential explanations for this, amino acid sequence differences in human versus rodent Aβ are known to limit the ability of rodent Aβ to form amyloid. 6

To overcome this technical obstacle, we developed a model of brain trauma in transgenic mice that develop AD-like Aβ plaques in a specific brain region beginning at 6 months of age. These mice were generated using a construct with the platelet-derived growth factor promoter driving a human β-amyloid precursor protein minigene containing the familial AD mutation V→F at APP position 717 (PDAPP). 10,11 In the present study, we evaluated the effects of brain trauma in the PDAPP mice at 4 months of age, ie, 2 months before the appearance of AD-like pathology, on memory function, histopathology, and regional tissue levels of Aβ peptides.

Materials and Methods

Brain Injury

To induce brain trauma, we used a recently described mouse model of rigid cortical indentation that dynamically deforms the left parietal cortex. 12 Briefly, brain trauma was induced by impacting a 3-mm diameter impounder onto the cortex (5 m/s, 1 mm depth) through a 5-mm craniectomy. We induced brain trauma or sham treatment (surgery without brain impact) in PDAPP neutered female mice at 4 months of age. As controls, we also induced brain injury in or sham treated 4-month-old neutered female wild-type littermates.

Memory Evaluation

The water maze paradigm has been described in detail. 12,13 Briefly, the water maze is a circular pool 1 m in diameter. PDAPP and wild-type mice were trained to swim to a plexiglas platform submerged 1 cm, which they found by navigating using external visual cues. The mice were given 20 to 25 trials over 2 to 3 days to reach criterion, the last trial being 1 hour before brain damage or sham treatment. Only those reaching criterion were used. The animals that could perform the task were subjected to brain trauma (n = 14 wild-type, n = 13 PDAPP) or sham treatment (n = 14 wild-type, n = 15 PDAPP), and evaluated 1 week later for their ability to recall the platform location. The platform was removed and the mice were given 1-minute probe trials in the water maze while a video computer recording unit tracked their swim path. A memory score was derived by determining the relative amount of time spent in or near the former platform site. Statistical analyses were performed using two-way analysis of variance for all groups followed by a posthoc t-test with Bonferroni correction for individual comparisons.

Histopathological Analysis

The mice were humanely euthanized with an overdose of sodium pentobarbital (200 mg/kg) and transcardially perfused with saline after brain injury or sham treatment at 1 week (n = 5 injured, n = 3 sham wild-type mice; n = 5 injured, n = 3 sham PDAPP mice), 2 weeks (n = 5 injured, n = 3 sham wild-type mice; n = 5 injured, n = 3 sham PDAPP mice) and 2 months (n = 8 injured, n = 3 sham wild-type mice; n = 8 injured, n = 3 sham PDAPP mice). The brains were removed and fixed in 70% ethanol, embedded in paraffin, and cut into 6-μm sections. Brain sections were immunostained with a variety of previously characterized monoclonal and polyclonal antibodies that are highly specific for distinctive epitopes within the Aβ peptides, including 2332, BA-27, and BC-05. 7,14-16 Alternate sections were stained with hematoxylin, eosin, and cresyl violet. Light microscopic examination was then performed on the sections. Because we found an extreme loss of hippocampal CA₂ and CA₃ neurons in the brain-injured PDAPP mice, we selected one region of the dorsal hippocampus to perform neuron cell counting. Three adjacent sections from bregma−1.7 17 were selected for each sham or injured animal sacrificed 1 to 2 weeks posttrauma (n = 10/injured group, n = 6/sham group). Stained neurons with clearly identifiable nuclei were manually counted via light microscopy in the CA₂ and CA₃ regions of the hippocampus ipsilateral to injury. These regions, selected according to atlas determination, 17 encompassed the entire CA₂ and the CA₃ from the margin of the CA₂ to the most lateral extension of the dentate gyrus. The average number of neurons from each region of the three sections for each animal were determined and used to determine group means. Statistical analysis of the neuron loss was performed using a two-way analysis of variance for all groups followed by a posthoc t-test with Bonferroni correction for individual comparisons.

Aβ Tissue Levels

PDAPP mice and wild-type littermates were euthanized (see above) at 0 hours (sham), 2 hours, 6 hours, 24 hours, 3 days, and 7 days posttrauma. Following sacrifice, the brains were rapidly removed and dissected on a chilled plate isolating cortex, hippocampus, thalamus, brainstem, and cerebellum. These samples were Dounce-homogenized in 6.7 volumes (w/v) of 70% formic acid at 4°C. Homogenates were centrifuged at 45,000 rpm, 4°C for 1 hour. The supernatant was diluted 1:20 in 1 mol/L Tris base. Samples (n = 4–12/group) were evaluated with sandwich enzyme-linked immunosorbent assays for Aβ peptides as previously described. 10,18 Absorbencies falling within the standard curve for each assay were converted to fmoles. Sandwich enzyme-linked immunosorbent assays for Aβ were prepared with the capture antibody BAN-50, a monoclonal antibody specific for amino acids 1–16 of Aβ, and reporting antibodies BA-27, specific for Aβ ending at Aβ40, and BC-05, specific for Aβ ending at Aβ 42.⁴³ The assay has a detection limit of <6 fmole/well for Aβ_1–40 and Aβ_1–42. The monoclonal antibodies BAN-50, BA-27, and BC-05 were prepared as described previously. 19 Statistical analyses were performed using a two-way analysis of variance for all groups followed by a posthoc t-test with Bonferroni correction for individual comparisons.

Results

Memory Evaluation

We found that although all wild-type mice easily learned the spatial memory task before injury, approximately 20% of the transgenic animals did not reach criterion and were excluded from further analysis. Of the animals that did reach criterion, several significant differences were found between groups evaluated one week later for memory retention of the task. Consistent with previous reports, wild-type brain-injured animals demonstrated significant memory dysfunction compared to their respective sham group (P < 0.001). 12 In addition, a significant impairment of memory retention was found in the sham PDAPP mice compared to sham wild-type animals (P < 0.01). Moreover, brain-injured PDAPP mice had a more profound impairment of memory function than did injured wild-type animals (P < 0.001) (Figure 1) ▶ . Importantly, the swimming speed of the mice in all groups was identical, demonstrating that potential motor impairment did not influence performance.

Figure 1.

Histogram demonstrating relative memory retention of a water maze task of wild-type and PDAPP mice one week following brain trauma or sham treatment. *P < 0.01, **P < 0.001.

Histopathological Analysis

Brain sections immunostained with a variety of antibodies that are highly specific for distinctive epitopes within the Aβ peptides revealed the presence of only a few Aβ-containing plaques, primarily in the hippocampus, that were indistinguishable between the 4-month-old brain-injured and sham-treated PDAPP animals. Although 6-month-old PDAPP animals demonstrated several more Aβ-containing plaques, consistent with previous reports, 10,11 there was no discernable difference in the number or distribution of plaques between brain-injured (2 months postinjury) and sham-treated animals. Thus, brain trauma did not accelerate or exacerbate Aβ deposition in the PDAPP mice.

Nonetheless, we did observe a dramatic difference in hippocampal pathology in the brain-injured PDAPP versus the wild-type mice. At both 1 and 2 weeks after injury, the wild-type mice demonstrated a modest loss of hippocampal neurons in a patchy distribution throughout the ipsilateral CA₂ and CA₃ subfield exactly as described earlier. 20 In striking contrast, the entire length of the CA₃ pyramidal cell layer of the brain-injured PDAPP mice degenerated completely with virtually no or very few neurons remaining (Figure 2) ▶ . Quantitative analysis of the CA₂ and CA₃ regions showed that the number of neurons in the hippocampi of wild-type and PDAPP sham animals were nearly the same. Brain injury in both the wild-type and PDAPP mice induced a highly significant loss of neurons in the CA₂ and CA₃ regions of the hippocampus compared to respective sham groups (P < 0.001). The wild-type animals had a 52% loss of CA₂ neurons compared to a 58% loss for the PDAPP animals, an insignificant difference. However, in the CA₃ region the wild-type brain-injured mice demonstrated a 36% loss of neurons whereas the PDAPP brain-injured mice showed a remarkable 84% loss, compared to respective sham groups. This difference between injured groups was found to be highly significant (P < 0.001). These results are shown in Figure 3 ▶ .

Figure 2.

Representative Nissl-stained coronal brain sections of the dorsal hippocampi of PDAPP and wild-type mice one week following sham treatment or brain trauma. A: Normal neuron density and orientation of the hippocampus in a PDAPP mouse. B: Typical loss of neurons in the CA₂ and CA₃ subfield of brain injured wild-type mice (small arrows), C-F: Hippocampi from four brain-injured PDAPP mice demonstrating an almost complete loss of CA₃ neurons (large arrow, c).

Figure 3.

Histogram demonstrating neuron cell loss in the CA₂ and CA₃ region of the hippocampus of brain injured wild-type and PDAPP mice. *P < 0.001 compared to respective sham group. ^†P < 0.001 compared to brain injured wild-type mice.

Aβ Tissue Levels

Aβ_1–40 and Aβ_1–42 levels in samples from both sham and brain-injured wild-type mice were either very low or below the level of detection at all timepoints. Thus, no injury effect on Aβ levels could be discerned in the wild-type mice. In the sham PDAPP mice, the highest levels of Aβ_1–40 and Aβ_1–42 were expressed in the hippocampus at twice the concentration found in the cortex, with the Aβ_1–42 present at levels similar to those previously reported in PDAPP mice at 4 months (Figure 4) ▶ . 10,11 In addition, very low amounts of Aβ_1–40 and Aβ_1–42 were found in the thalamus, cerebellum, and brainstem in the sham PDAPP mice. Two hours after trauma in PDAPP mice, a dramatic sevenfold increase in the concentration of Aβ_1–42 and threefold increase in Aβ_1–40 were observed in hippocampal tissue (P < 0.001), whereas smaller but significant increases were seen in cortical tissue (P < 0.01). Thus, the ratio of the levels of Aβ_1–40 compared to Aβ_1–42 also changed in the hippocampus after injury. In sham animals Aβ_1–42 was found at three times the concentration of Aβ_1–40, but at 2 hours posttrauma, the concentration of Aβ_1–42 became 10-fold greater than Aβ_1–40. However, by 6 hours posttrauma, Aβ_1–40 and Aβ_1–42 concentrations in the hippocampus had returned to baseline values and by 24 hours, Aβ_1–42 levels fell below baseline in the cortex (Figure 4) ▶ . No other regions demonstrated detectable changes in Aβ concentrations after brain injury in PDAPP mice.

Figure 4.

Histogram demonstrating the temporal changes in the concentration of Aβ _1–40 and Aβ_1–42 following brain trauma in PDAPP mice in the hippocampus and cortex. *P < 0.01, **P < 0.001 compared to respective sham Aβ levels. Data are not shown for wild-type mice because their Aβ levels were at the minimal detection limit (6 fmol/well) or below for all time points.

Discussion

These results demonstrate that brain trauma in PDAPP mice induces massive neuron death selectively in the hippocampus, the extent of which has never been observed following the same or greater levels of brain trauma in wild-type mice. 12 Additionally, acute and marked increases in Aβ peptide levels were also found selectively in the hippocampi of PDAPP animals after brain trauma. However, despite these posttraumatic increases in Aβ brain trauma did not accelerate or augment Aβ plaque formation 2 months after injury in PDAPP mice. In this study we also found that PDAPP mice demonstrate memory dysfunction before the appearance of overt AD-like pathology, consistent with previous reports using other transgenic mouse strains. 20,21 Although there appeared to be an exacerbation of posttraumatic memory dysfunction in the PDAPP mice compared to wild-type littermates, the nature of this difference could not be determined due to the baseline deficit in the noninjured PDAPP animals.

The acute posttraumatic increase of Aβ_1–42 in the 4-month-old PDAPP mice reached levels similar to those found in 1-year-old noninjured PDAPP mice when substantial Aβ deposition is observed throughout the hippocampus. 10,11 Notably, the high levels of Aβ in the brain-injured PDAPP mice were associated not with plaque formation but with neuron death. However, hippocampal neuron death has never been reported in the absence of brain injury in this or another line of transgenic mice that overexpress mutant β-amyloid precursor proteins and develop increasing numbers of AD-like amyloid plaques with advancing age 22,23 Moreover, no induction of neuron death has been seen after Aβ injections in rat brains. 24,25 In association, these findings suggest that in vivo Aβ may not be neurotoxic unless a second stress or insult triggers Aβ to induce neuron death. In support of this hypothesis, a very recent report demonstrated increased susceptibility to ischemic brain damage in another line of transgenic mice overexpressing mutant APP. 26 Although there was no difference in Aβ levels between ischemic and nonischemic hemispheres at 24 hours postinjury, high baseline levels of Aβ were present in the brains of these transgenic animals. Thus, we propose a “two-hit” hypothesis for AD wherein Aβ is necessary but not sufficient to cause neuron death until a second pathological process potentiates Aβ neurotoxicity.

Because an acute and marked posttraumatic increase in the concentrations of both Aβ_1–40 and Aβ_1–42 implicates Aβ peptides as potential deleterious agents in the posttraumatic sequelae, these findings suggest a mechanistic link in the pathogenesis of brain trauma and AD which may have important clinical implications. Accordingly, recent reports have shown that brain trauma patients with apolipoprotein E-4 genotypes, a genetic risk factor for AD, have much worse outcomes after brain trauma, even in the very acute setting. 27,28,29 Taken together, these findings suggest that the role of brain trauma in AD can be explored in greater detail using the model presented here.