Traumatic Brain Injury Alters the Trajectory of Age-Related Mitochondrial Change

Mice and procedural overview

The University of Kansas Medical Center Institutional Animal Care and Use Committee (Animal Care and Use Protocol #2018-2475), and the USAMRMC Animal Care and Use Review Office (ACURO) (protocol AZ170111.e001) approved all experimental animal protocols and procedures. The number of mice used for these studies was based on power calculations and mortality estimates, in which we anticipated a 30% attrition rate for head-injured mice maintained over the entire duration of the study.

We purchased 76 male, 4-month-old C57BL/6J mice and 23 male, 4-month-old rTg4510 mice from the Jackson Laboratory. The mice were acclimated to our vivarium for 3 weeks, group-housed to a maximum of 5 littermates per cage in Touch Slim Line IVC cages (Techniplast) and maintained on a 12-h light/dark cycle. Some mice were separated on an individual basis upon recommendation of the institutional veterinary staff for excessive fighting; singly housed mice were provided additional nestlets for environmental enrichment.

The non-transgenic (non-tg) mice throughout received a standard chow diet (LabDiet product # 5053) with ad libitum access to food and water. The rTg4510 mice, which express a repressible form of human tau that contains a P301L mutation that drives intraneuronal tau fibril accumulation [18], were maintained until 12 months of age on a doxycycline-enriched (200 mg/kg doxycycline hyclate) chow diet to repress transgene expression. After that point, tau transgene expression was activated by switching these mice to the standard chow diet. The rTg4510 mice also had ad libitum access to food and water.

At 5 months of age the young adult, non-tg mice were randomized to receive either a controlled cortical impact (CCI) or sham surgery. All the 5-month-old rTg4510 mice received a CCI surgery. The non-tg mice were euthanized at either 6 (n = 29; 14 shams) or 20 (n = 44; 21 shams) months of age. We initially intended to age the tg mice to 20 months, but that cohort experienced higher than expected age-dependent mortality and the remaining mice (n = 15) were euthanized at age 17.5 months.

Euthanasia was performed by isoflurane overdose, at which point the brains from the non-tg mice were rapidly removed and dissected. The parietal cortex and hippocampal regions were separated, designated as either ipsilesional or contralesional to CCI/sham, and were placed in phosphate buffered saline (PBS) and immediately frozen at –80°C. The frontal lobes from the transgenic mice were formalin-fixed, embedded in paraffin, sectioned, mounted on slides, and stored at room temperature.

The parietal cortical regions were used for electron transport chain (ETC) assays, the hippocampi for mitochondrial DNA copy number (mtDNAcn) determinations, and the formalin-fixed frontal tissue for histologic analyses (described below). Performing biochemical and molecular assays on frozen rather than fresh tissue allowed us to analyze all the samples together and thus avoid batch-effect variation. The mice were also characterized through behavioral testing and brain magnetic resonance imaging (MRI), per protocols and schedules as described. All assessments of living non-tg mice, or on their collected tissues, were performed in a blinded fashion. Figure 1 summarizes the study design and timeline.

OPEN IN VIEWER

Fig. 1

Overview of study flow and procedures. A) Timeline for the non-tg mouse experiments. B) Timeline for the tg mouse experiments.

CCI and sham surgeries

A CCI targeting the right sensorimotor cortex was produced using a commercial impactor (Impact One Stereotaxic CCI Instrument, Leica Microsystems) as described in our previous publications [19, 20]. Briefly, the device was mounted on a stereotaxic frame to reproduce the position and direction of the impact with high precision. Following anesthesia with isoflurane (3% induction, 1–2% maintenance) mice were immobilized in the stereotaxic frame and placed on a feedback-controlled heating pad to maintain core body temperature.

All surgeries were carried out by a single surgeon under aseptic conditions. The scalp was retracted, and a 4 mm diameter circular craniectomy was performed with a trephine bit (Patterson Dental), centered 0.5 mm anterior and –2.0 mm lateral to bregma. The impactor tip, which consisted of a 3 mm diameter stainless steel rod, was centered within the craniectomy and angled so that the face of the impactor tip was tangential to the dural surface. The impactor tip was slowly lowered until contact with the dura was indicated by a contact sensor alarm. The tip was then retracted, and the trigger switch activated to deliver the cortical impact at 3.5 m/s, 2.0 mm depth, and 100 ms contact time. The cranial defect was repaired with a sterilized plastic cap secured with Vetbond adhesive. The incision was closed with 4-0 Vicryl sutures and treated with a topical analgesic (EMLA cream) and antibiotic ointment. Mice were removed from the stereotaxic frame and placed in a heated recovery chamber. The sham-injured control mice received an identical craniotomy, but the bone flap was left in place to avoid meningeal damage. The anesthesia duration for the sham surgeries was matched to that of the CCI surgeries.

Magnetic resonance imaging

MRI scans were obtained at 9.4T on 1–2% isoflurane-anesthetized mice maintained at a respiration rate of 100–150/min; we used a feedback-controlled heating pad to keep core body temperature at 37°C. Gradient echo multi-slice images (TR = 65 ms, TE = 2.8 ms) were used to position the mouse’s head in the magnet isocenter, followed by rapid acquisition with relaxation enhancement T₂-weighted images (TR = 4000 ms, TE = 0.01 ms, echo train length = 16, slices = 17, slice thickness = 0.5 mm). T₂-weighted images were analyzed with a semi-automated MatLab program developed in-house to quantify the TBI lesion using a modified Cavalieri approach.

Brain tissue loss was calculated by subtracting the contralesional hemisphere cerebrospinal fluid (CSF) volume from the ipsilesional hemisphere CSF volume in mm³. To account for variance in brain volume between mice and within individual mice over the course of aging, we normalized the lesion to total brain volume in mm³ obtained from the MR images. We compared sham, CCI, and tg-CCI mice at 1-month post injury and compared the change in brain lesion from 1-month to 15-months post-injury.

Behavioral assessments

We only performed behavioral assessments on the non-tg mice. The test battery included two motor tasks (rotarod, grid walk) and one cognitive task (Barnes maze). All behavioral assessments were conducted in the morning (8 : 00–11 : 00 am) and in the same order (Barnes, grid walk, rotarod) across testing days. At the start of the study, mice were acclimated to the behavioral apparatus for one day, then trained on the tasks for three days (days –5 to –3) to establish a reliable pre-injury performance baseline. After administering CCI or sham surgery, behavior was tested at an “acute” stage (days + 1 to +4), a “recovery” stage (1-month post), and a “post-aging” stage (15-months post). A detailed description of the behavioral testing methodology can be found in the Supplementary Methods section.

Electron transport chain V _max assays

We used methods previously established by our group [21, 22] to quantify mitochondrial complex I and complex IV (cytochrome oxidase; COX) Vmax enzyme activities. The assays were performed on homogenates of parietal cortex tissue from the non-tg mice. We specifically evaluated both parietal cortices from the CCI-injured mice, and the ipsilesional parietal cortex from the sham-injured mice at 1- and 15-months post CCI/sham. The amount of total protein present in each homogenate was determined using a BCA protein assay kit (BioRad, Hercules, CA), and the spectrophotometrically determined Vmax rate for each sample was normalized to its corresponding protein value. Readings were obtained using an Infinite M200 plate reader (Tecan). We performed an additional calculation in which we normalized each sample’s Vmax/protein activity to the hippocampal mtDNAcn mean value for its corresponding sample group, or to the mtDNAcn from its corresponding hippocampal tissue (see below).

Mitochondrial DNA copy number determination

We performed mtDNAcn measurements on whole hippocampi. From the non-tg and rTg4510 mice that received a CCI we obtained both ipsilesional and contralesional data. From the non-tg sham mice we obtained only ipsilesional data. Our mtDNAcn sample sizes are slightly smaller than our ETC sample sizes as 2–3 hippocampal samples from each group were directed towards other applications, and one sample (in the sham 15-month group) could not be confidently identified. We estimated mtDNA to 18 S rRNA DNA ratios following reverse transcription PCR (RT-PCR) using Promega GoTaq G2 Hot Start Master Mixes Colorless, Syto-82 (Thermo Fisher Scientific) 5 mM, 4μl/mL, and USB ROX Passive Reference Dye (Affymetrix 75768) 4μl/mL [23]. A detailed description of the mtDNAcn determination methodology can be found in the Supplementary Methods section.

Histopathology

Frontal cortices from nine of the 15 rTg4510 mice were suitable for tau histologic analysis, and five had enough residual tissue to also permit NeuN and glial fibrillary acid protein (GFAP) staining. We isolated the frontal cortex with a vertical cut at Bregma±1.2 mm then fixed with 4% paraformaldehyde in PBS for 4 days at 4°C. After fixing, the brain was embedded in paraffin wax using a Leica Biosystems ASP6025 S tissue processor. The paraffin blocks were cut into 7μm thick sections and mounted directly on Superfrost/Plus microscope slides (Fisher, # 12-550-15). The slides were dried at 40°C overnight and stored in slide boxes at room temperature until further processing. Methodological details related to tissue processing, staining, and stereological analysis can be found in Supplementary Methods.

Statistics

The outcomes were summarized with respect to mean and standard error of the mean (SEM) for each defined group and bar plots of the means with error bars. Differences in outcomes between two independent groups of mice were assessed using two sample t-tests, and differences in outcomes from the same set of mice were assessed using paired t-tests. Comparisons of outcomes among three groups were performed using one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test. The differences in the behavioral outcomes (rotarod, grid walk, Barnes maze) between the groups that were measured over time were assessed using a mixed effects modeling framework for repeated measures data using SAS procedure GLIMMIX. Spearman’s rank correlation analyses were carried out to examine relationships between outcomes. p-values<0.05 were considered statistically significant.

CCI produced brain lesions that evolved during aging

Long-term survival of the non-tg mice after injury was better than we predicted, while survival of the transgenic mice was worse than predicted. One mouse in the 1-month sham-injured group died following surgery; all the 1-month CCI mice survived. During the aging period, two non-tg mice from the 15-month sham-injured group died and did not contribute mtDNAcn or ETC data. All the CCI-injured mice survived the aging period. Eight out of the 23 transgenic mice died during the aging period (34.8%), and most of these deaths (6 out of 8) occurred during the period following removal of doxycyline at 12 months of age. Supplementary Figure 1 shows the number of mice that contributed to each component of the study.

We administered CCI or sham injuries to the right sensorimotor cortex of 5-month-old C57BL/6J and tau transgenic rTg4510 mice. One month later, MRI assessments showed a cystic brain lesion below the site of impact with ipsilesional hemisphere volume loss in the CCI mice, and essentially no observable structural injury in the sham mice (Fig. 2A, B). Mean absolute lesion volumes at 1 month were 0.24±0.09 mm³ in the sham group, 5.34±3.16 mm³ in the CCI group, and 7.19±3.51 mm³ in the tg-CCI group. Lesion volume normalized to total brain volume was larger in non-tg CCI group than in sham (p < 0.0001), and larger in the tg CCI group than in non-tg mice (p = 0.005) even though the tau transgene expression was suppressed by doxycycline during this period (Fig. 2C).

OPEN IN VIEWER

Fig. 2

Brain structure and volume changes. A) A representative MRI taken 1 month after sham surgery shows the sham procedure did not grossly alter brain structure. B) A representative MRI taken 1 month after CCI shows a cystic lesion underlying the region of impact. C) Brain lesions were quantified 1-month post-injury as the volume of the contralesional hemisphere CSF subtracted from the volume of the ipsilesional hemisphere CSF and normalized to total brain volume. Normalized lesions were larger in the tg CCI group than in non-tg mice (p < 0.05) even though the tau transgene expression was suppressed by doxycycline during this period. D) Over the course of the 14-month aging period the brain lesions continued to expand, increasing much more in non-tg CCI mice than in the sham-injured mice. E) Increasing frailty of the tg mice, reflected here by premature weight loss, prompted us to forego post-aging MRI in the tg cohort.

Over the 14-month aging period the normalized lesion volume increased by 73% in the sham-injured group. This change was statistically significant (p = 0.021) although the area of injury remained very small in sham mice. Over the 14-month aging period the normalized lesion volume increased by 104% (p < 0.001) in the non-tg CCI group (Fig. 2D).

When the rTg4510 mice were 12 months old we removed the doxycycline from their chow to enable tau transgene expression. Around this time the transgenics began to lose weight and became increasingly frail (Fig. 2E). For this reason, we did not obtain post-aging follow-up MRIs in the transgenic cohort.

CCI and aging affected behavioral testing

The CCI affected rotarod test performance. In non-tg mice, rotarod performance 1–4 days post-CCI was markedly impaired relative to baseline and to the sham mice. At the 1-month time-point the CCI group rotarod performance improved, although not to the pre-injury level or to the level of the sham group. At the 15-month time-point the CCI group’s rotarod performance was still slightly impaired. The relative difference between CCI and sham group performance at the 15-month timepoint was comparable to the relative amount of impairment at the 1-month timepoint. The sham group showed consistent rotarod performance across all measurements. (Fig. 3A). In both groups, for unclear reasons rotarod performance on the first day of the post-aging evaluation differed from that of the subsequent testing days, but this aberration did not alter the outcome of the statistical analysis.

OPEN IN VIEWER

Fig. 3

Behavioral testing. A) On the rotarod test, the CCI group showed impaired performance in the immediate post-injury period (p < 0.0001). Their performance improved over the next month but did not return to the level of the sham group (p = 0.006). Fifteen months post-injury the CCI mice rotarod performance did not show further change (p = 0.010). The sham group showed consistent rotarod performance across all measurement periods. B) On the grid walk test, relative to sham-injury, a CCI injury increased the number of left foot faults normalized to the total walking time. This deficit persisted through the acute (p < 0.0001), 1-month (p < 0.0001), and 15-months (p < 0.0001) post injury timepoints. C) In the Barnes maze, latency to goal for the CCI group was prolonged in the immediate post-injury period (p = 0.010) but recovered over the next month to match the sham group time (p = 0.702). The latency to goal time did increase over the subsequent 14-month aging period, although this change was comparable between the CCI and sham groups (p = 0.509). D) The performance pattern for distance traveled on the Barnes maze reflected the pattern shown in (C) (p = 0.033 immediate post injury, p = 0.444 at one month, and p = 0.165 at 14 months). E) During the Barnes memory probe trial when the escape box was removed, sham and CCI groups showed similar latency to goal perimeter scores at all time points. p-values <0.05, <0.01, <0.001, <0.0001 are represented by ^*, ^**, ^***, and ^**** respectively. The vertical lines represent mean difference between groups. Horizontal solid (dashed) lines represent mean difference in TBI (Sham) between the phases.

The CCI affected grid walk test performance. At all post-injury timepoints, the number of contralesional foot faults normalized to the total walking time was higher in CCI mice than in the sham-injured mice. The CCI-injured mice improved over time, performing better at 1 month than they did immediately after injury, and better again at 15-months post. With aging there was a non-significant trend towards worsening performance in the sham group (p = 0.168), but overall, the sham group performance remained consistent throughout the study (Fig. 3B).

The CCI affected performance on the Barnes maze acutely after injury. During the immediate post-injury period the distance traveled and latency to goal scores were worse in the CCI group than they were in the sham group (p = 0.033 and p = 0.010); the extent to which motor dysfunction contributed to this is unclear. These deficits were no longer apparent at the 1-month post-injury time-point. At 15-months post-injury the distance traveled and latency to goal scores were not as good as they were at the 1-month time-point, but the extent of change with aging was comparable between the CCI and sham groups (Fig. 3C, D). During the Barnes probe trial when the escape hatch was removed, CCI and sham mice had similar latency to goal scores acutely, 1-month, and 15-months post-injury (Fig. 3E).

CCI accelerated the trajectory of age-related mitochondrial change

To determine whether TBI affected the trajectory of aging-related mitochondrial changes, we measured complex I and COX Vmax activities in the parietal cortex from CCI and sham non-tg mice. These assays were performed using brain homogenates in which we normalized enzyme-driven, time-dependent spectrophotometric changes to total protein. One month post-injury, complex I and COX activities were equivalent among the sham contralesional, CCI contralesional, and CCI ipsilesional tissues (Fig. 4A, B); the CCI ipsilesional and CCI contralesional values remained comparable by paired t-test analysis. The complex I and COX activities were also equivalent among all three groups at the 15-month post-injury time-point (Fig. 4C, D); the CCI ipsilesional and CCI contralesional values remained comparable by paired t-test analysis. Finally, we considered whether aging affected the enzyme activities. At 15-months post-injury, the complex I activity in the ipsilesional, injured hemispheres of the CCI mice was lower than it was at 1 month (21% reduction, p = 0.036) (Fig. 4E). We did not observe any other differences between the 1-month and 15-month enzyme activity means (Fig. 4F–H).

OPEN IN VIEWER

Fig. 4

Electron transport chain activities. Complex I and COX Vmax activities were assessed in parietal cortex. A) One-month post-injury, complex I activities were equivalent among the sham, CCI-ipsilesional, and CCI-contralesional groups. B) COX activities were equivalent 1-month post-injury. C) Complex I activities were equivalent 15-months post-injury. D) COX activities were equivalent 15-months post-injury. E) The complex I activity in the CCI-ipsilesional (injured) cortex was lower at 15-months than it was 1-month post-injury. F) The complex I activities in the CCI-contralesional cortex were equivalent at 1 and 15 months. G) The COX activities in the CCI-ipsilesional cortex were equivalent at 1 and 15 months. (H) The COX activities in the CCI-contralesional cortex were equivalent at 1 and 15 months.

To further assess the impact of TBI on the trajectory of age-related mitochondrial changes, we measured mtDNAcn in the hippocampi from CCI and sham non-tg mice. One month post-injury, mtDNAcn values were equivalent among the sham ipsilesional, CCI ipsilesional, and CCI contralesional hippocampi (Fig. 5A). Over the next 14 months the mtDNAcn increased in all groups, and in the old mice the mtDNAcn of the CCI contralesional samples exceeded that of the CCI ipsilesional and sham samples (Fig. 5A). We further characterized the extent of each group’s age-related increase as a percentage change over 14 months. In this analysis there was a non-significant trend towards a greater age-related increase in the CCI ipsilesional group versus sham (42% increase in sham brains versus 57% increase in CCI-ipsilesional; p = 0.136). The age-related increase in the CCI-contralesional group (78% increase) significantly exceeded that of the sham group (p = 0.002) and that of the CCI-ipsilesional group (p < 0.001) (Fig. 5B).

OPEN IN VIEWER

Fig. 5

Mitochondrial DNA copy number. A) One-month post-injury the sham ipsilesional, CCI-injured ipsilesional, and CCI-injured contralesional hippocampi showed a comparable mtDNAcn. The mtDNAcn for each group increased over the subsequent 14-month aging period. At 15-months post-injury the mtDNAcn in the CCI-contralesional group exceeded that of the sham-injured and CCI ipsilesional groups. The Y axis values are mtDNAcn to nuclear DNA copy number ratios. B) The extent of age-related mtDNAcn change, shown here as a percent increase of each 15-month measure compared to the mean group value at 1 month, was greater in the CCI-contralesional group than it was in the other two groups. C) The trajectory of each group’s mtDNAcn age-related change is depicted. Although mtDNAcn levels were comparable 1-month post-injury, 14 months later the mtDNAcn in CCI-contralesional hippocampus exceeded that in sham hippocampus (p = 0.029 by two sample t-test) as well as that in CCI-ipsilesional hippocampus (p = 0.009 by paired t-test). The Y axis values are mtDNAcn to nuclear DNA copy number ratios.

Figure 5C emphasizes the age-dependent trajectory of the hippocampal mtDNAcn changes in the different groups. A TBI administered during young adulthood appeared to accelerate an age-related increase in hippocampal mtDNAcn, although this acceleration was not as robust on the injured side of the brain as it was on the contralesional side.

To add context to the question of how aging affected our TBI model we more closely examined the effects of aging itself in our sham control group. In the sham-injured mice, we jointly analyzed the ipsilesional hippocampus mtDNAcn and contralesional parietal cortex enzyme activity measurements presented in Figs. 4 and 5. Between 6 and 20 months of age, the mtDNAcn in the brains of these mice increased by 42%. Their mtDNAcn, when normalized to nuclear DNA copy number, increased from a value of 31 to a value of 44 (Fig. 6A), while complex I and COX activities, when normalized to total protein, did not change (Fig. 6B, C). Further normalizing the complex I/protein and COX/protein enzyme activities to the mtDNAcn means, though, resulted in robust age-related differences (Fig. 6B, C). Specifically, the complex I/protein/mtDNAcn value was 34% lower at 20 months of age than it was at 6 months, and the COX/protein/mtDNAcn value was 37% lower at 20 months of age than it was at 6 months. For each brain where we had both an ETC Vmax and mtDNAcn value (n = 12 for the young mice; n = 17 for the old mice) we also normalized the Vmax activity to the corresponding mtDNAcn value, and again the mtDNAcn-referenced activities were lower in the old mice (Fig. 6D, E). This suggests that as the mice aged, to maintain brain complex I and COX activity homeostasis they needed to increase their brain mtDNAcn.

OPEN IN VIEWER

Fig. 6

Sham-injured mouse ETC activities normalized to mtDNAcn. A) Between 6 and 20 months of age, the mtDNAcn in the brains of the sham-injured mice increased by 42%. B) While the complex I activity, when referenced to total protein, did not change between 6 and 20 months of age, normalizing the complex I/protein activity from the 6 and 20-month-old mice to the corresponding 6 and 20-month-old mtDNAcn group means reveals a robust age-related difference. The Y-axis shows relative activities. C) While the COX activity, when referenced to total protein, did not change between 6 and 20 months of age, normalizing the COX/protein activity from the 6 and 20-month-old mice to the corresponding 6 and 20-month-old mtDNAcn group means reveals a robust age-related difference. The Y-axis shows relative activities. D) After normalizing each individual brain’s complex I/protein activity to its own mtDNAcn, the mean of the values for the 20-month-old mice (n = 17) was lower than it was in the 6-month-old mice (n = 12). E) After normalizing each individual brain’s COX/protein activity to its own mtDNAcn, the mean of the values for the 20-month-old mice (n = 17) was lower than it was in the 6-month-old mice (n = 12).

CCI-injured tau transgenic mice showed ipsi-versus contralesional differences in mtDNA copy number and tangle counts

We intended to age the tau tg rTg4510 mice for 15 months after CCI and then assess mtDNAcn and tau tangle frequency, but the cohort exhibited progressive mortality causing us to end the study at 12.5 months post-injury (age 17.5 months). At this post-aging time point, mean mtDNAcn in the contralesional hippocampus was 16% higher than on the ipsilesional side (34.3 for contralesional versus 29.6 for ipsilesional; p = 0.015 by paired t-test) (Fig. 7A). There was a positive correlation between the ipsilesional and contralesional mtDNAcn values from the individual mice (Supplementary Figure 2).

OPEN IN VIEWER

Fig. 7

Tg mice mtDNAcn and histopathology. A) For the 17.5-month-old rTg4510 mice that received a TBI in young adulthood, the mtDNAcn was higher in the contralesional hippocampus than it was in the ipsilesional hippocampus (p < 0.05 by paired t-test). B) Frontal cortex neurofibrillary tangles revealed through thioflavin fluorescence and immunofluorescence with the S396 antibody that binds phosphorylated tau. C) For both hemispheres the number of tangle-bearing cells was greater with thioflavin than with the S396 antibody, and there were cells that were positive for one or both stains. The number of thioflavin, S396, or co-stained cells was consistently greater in the contralesional hemisphere. D) The number of DAPI-NeuN positive cells, the number of DAPI-GFAP positive cells, the NeuN fluorescence intensity values, and the GFAP fluorescence intensity values were comparable between the ipsilesional and contralesional frontal cortices. Each parameter is expressed as a relative value of the ipsilesional cortex normalized to contralesional cortex.

In the rTg4510 mice 12.5 months post-injury, we identified frontal cortex neurofibrillary tangles through thioflavin fluorescence and immunofluorescence with the S396 antibody that binds phosphorylated tau. For both hemispheres the number of tangle-bearing cells was greater with thioflavin than with the P396 antibody, and there were cells that were positive for one or both stains (Fig. 7B, C). The number of thioflavin, P396, or co-stained cells was consistently greater in the frontal cortex contralesional to CCI (Fig. 7C). We did not detect significant correlations between mtDNAcn and tangle counts (Supplementary Figure 2).

Five of the transgenic mouse frontal lobe blocks contained adequate residual tissue for neuronal and glial assessments, and in these we counted the number of DAPI-positive cells with clear concomitant NeuN or GFAP staining. We also measured the ipsilesional and contralesional NeuN and GFAP fluorescence intensities. We did not observe inter-hemispheral differences in neuron number, glial number, NeuN intensity, or GFAP intensity (Fig. 7D).