Introduction

Head trauma remains a significant pediatric health problem, with an estimated incidence of 230/100,000 leading to 100,000–200,000 new pediatric traumatic brain injuries (TBI) each year, of which children 0–4 have the highest rates of TBI-related emergency room visits, hospitalizations, and deaths [1–5]. Approximately 10–15 % of childhood TBI is severe and leading to death or permanent brain damage [6, 7]. Significant motor and cognitive deficits can occur after mild to moderate TBI [8–11]. In fact, children less than 4 years old have a worse outcome in comparison to older children and adults [10, 12]. Hence, juvenile TBI burdens both families and society with health care and associated costs [5], for which outcome depends on age-at-injury and necessitates translational relevant knowledge to minimize adverse consequences of injury.

Controlled cortical impact (CCI) is an experimental model of TBI that is clinically relevant to brain injuries resulting in contusion. Primarily a focal cortical injury depending on depth of penetration, CCI allows for control over mechanical parameters (velocity, depth of impact), has been well studied, [13–20], and results in reproducible behavioral deficits in the developing, immature rat [21]. For developing animals, protocols from the adult have been adapted and scaled to study injury-induced pathology and behavioral deficits [21, 22]. Post-natal rats mature rapidly, such that by 28 days post-injury, they have reached adolescence or young adulthood. Unilateral CCI in the juvenile rodent permits histological analysis of both neuronal morphology in proximity (ipsilateral) and distal (contralateral) to the site of contusive injury.

Over 28 days after CCI in juvenile rodents, both transient and long-lasting behavioral deficits have been associated with cortical and hippocampal pathology. Rats receiving a 2.0-mm deflection CCI on post-natal day (PND) 17 showed significant memory and learning deficits in the Morris Water Maze paradigm beginning at post-injury day (PID) 11 and lasting until at least PID17 in comparison to uninjured sham rats [21]. In comparable studies using similar injury parameters, cognitive deficits resolved by PID28 [21, 23, 24]. However, somatic and emotional dysfunction, measured by sensorimotor function and anxiety-like behaviors, persist through PID60 after experimental TBI at PND17 in the absence of spatial learning and memory deficits [24]. Performance in these tasks depends on hippocampal function, suggesting that circuit reorganization and underlying changes in neuron morphology may mediate the transient deficits and long-lasting deficits.

Long-lasting changes in neuronal structure occur after CCI in PND17 rats, where a greater number of neurons have incorporated into hippocampal circuitry at PID28 [25]. Injury-induced damage that extends into white matter tracts at PID60 suggests that TBI at a young age can cause long-term connectivity changes alongside persisting behavioral deficits [24]. In impact-acceleration TBI at PND17, cognitive dysfunction endures through 3 months post-injury [26], despite no obvious neuronal loss [27]. Although injury-induced alterations in circuit structure have been implicated in both transient and long-lasting behavioral deficits, cytoarchitectural changes have not been evaluated at the level of individual neurons. In addition, the developmental period at the time of injury impacts outcome. PND17 is likely beyond the most vulnerable developmental period, but histopathological analysis of the immature brain after impact-acceleration brain trauma revealed peak apoptotic cell death up to PND14, and vulnerability rapidly decreased thereafter [28, 29]. Thus, age-at-injury may influence outcome in regard to neuronal circuitry, with consequences for functional outcome.

In the present study, we sought to further quantify the morphological changes that occur concurrent with development and reorganization of the hippocampus at 28 days post-injury, building on our previous studies [21, 25, 30]. Our goal is to regionally define morphological changes in hippocampal neurons to attribute circuit alterations to the nature of cognitive deficits previously reported [21]. The Golgi method is ideally suited for defining trauma-induced pathology because it reveals the details of cellular morphology that cannot be readily assessed with other histological methods. The extent of these changes will be measured by analyzing soma area, dendritic length, and dendritic branching. We hypothesize that CCI in the juvenile rat will lead to chronic quantifiable structural changes in neurons of hippocampal subregions. Furthermore, there will be injury-induced changes in morphology in younger (PND7) animals that will differ from PND17.

Methods

Experimental groups

Adult female Sprague-Dawley rats with timed pregnancy (Harlan Sprague-Dawley; Indianapolis, IN) were used with each litter culled to eight male pups. At PND17, the rat pups from the same litter were randomly assigned to one of three groups: trauma (n = 4), sham (n = 2), and naïve (n = 2). A second litter at PND7 was assigned to one of two groups: trauma (n = 6) and sham (n = 2). The dams and their litters pre- and post-injury all followed the same protocol and were maintained in standardized conditions throughout the study (12-h light/dark cycle with lights on at 0700; 22 °C). Food and water were available ad libitum. These experiments were approved by the University of Pittsburgh Institutional Animal Care and Use Committee (protocol #0009975).

Surgical preparation

The surgery and injury have previously been described [21]. Briefly, on PND17 (or PND7), rat pups were initially anesthetized with 1.5–2.0 % isoflurane in N2O/O2 (2:1) using a nose cone; core temperature was maintained at 37.0–37.5 °C throughout the procedure by a heating pad and monitored by rectal thermometer. The scalp was shaved and sterilely prepared. The head was secured in a stereotaxic head frame and the skull exposed by a midsagittal incision and reflection of the scalp. Using a high-speed air drill, a craniotomy (7.0 mm for PND17; 4.0 mm for PND7) was made in the left parietal bone at 0.2 mm posterior to the coronal suture and 0.2 mm anterior to lambdoid suture, extending laterally to the superior temporal line.

Controlled cortical impact

Controlled cortical impact (CCI) was performed to induce severe traumatic brain injury, as described [21, 25, 31]. Briefly, the dura mater was exposed by removing the bone flap, and the piston of the CCI was positioned to deliver a focal cortical impact. Anesthesia was discontinued until the toe pinch reflex returned. The impact was delivered with a 6-mm or 3-mm tip (PND17 or 7, respectively) at a controlled velocity of 4 m/s over 500 ms. The piston was calibrated to produce an injury deflection depth of 2.0 mm for PND17 animals. In PND7, the injury depth was reduced to 1.75 mm to account for the smaller size of the animals, as previously described [21]. Sham controls were subjected to the same surgical procedures with no impact. Naïve animals had neither surgery nor injury. Immediately following injury, anesthesia was resumed, the bone flap replaced and secured with Koldmount dental cement (Vernon-Benshoff, Albany, NY), and the scalp incision sutured. The rats were transferred to a warmed recovery cage and observed for at least 30 min. After resumption of normal gait (~30 min), the rat pups were returned to their litters and observed intermittently over the next 30 min to ensure acceptance by their dams.

Golgi-Cox procedure

On post-injury day 28 (PID28), all animals were anesthetized and perfused transcardially with a 10 % formalin solution using a peristaltic pump. Each brain was removed, blocked in 6-mm coronal sections, and post-fixed for 3 days at room temperature. Using a modification of the Golgi-Cox procedure that produces extensive and reproducible impregnation of neurons, the tissue sections were transferred to a Golgi-Cox solution that consists of 20 parts 5 % potassium dichromate, 20 parts 5 % mercuric chloride, 8 parts 5 % potassium chromate, and 40 parts distilled water. The tissue was kept in solution in the dark at room temperature for 2 months, with a transfer to a fresh solution each week [32, 33]. This period of incubation was optimized to maximize the silver impregnation of the neurons and their branches and dendritic arbors while minimizing impregnation of overlapping neurons. After this incubation period, the tissue blocks were sectioned serially through the rostrocaudal extent of the injury in the coronal plane using a vibratome at 200 μm and collected sequentially in 0.1 M sodium phosphate buffer. Thereafter, sections were dehydrated in a graded ethanol series (50, 70, 95, 100 %), cleared in clove oil, infiltrated in Cytoseal 60 (VWR Scientific), and coverslipped on gelatin-coated slides using Cytoseal 60.

Data collection/analysis

Data collected from the injured hippocampus were compared to homotypic regions of the contralateral uninjured hemisphere and from equivalent regions of controls (sham and naïve animals for PND17). Using Neurolucida software (MicroBrightField, Inc., Colchester, VT), the location of all impregnated perikarya was mapped for granule cells of the DG and pyramidal cells of area CA3 and area CA1. Neurons were selected for reconstruction based on the presence of an intact soma, the absence of truncations of large dendrites, and presence of natural ends. A natural end was defined as the termination of a completely tapered dendrite, identifiable by the presence of an end swelling, spine or spine cluster, or gradual narrowing. 3D reconstruction of neuron morphology was carried out by an investigator blinded to injury status using Neurolucida. The boundaries of the soma of each neuron were traced using a ×60 objective, and the predicted contour of the cell soma was followed. Then, all dendrites emerging from the cell soma were traced through their full extent. The completed reconstructions were analyzed using the NeuroExplorer program (MicroBrightField, Inc., Colchester, VT) to calculate soma area, dendritic length, and branch points. The linear extent of all components of the dendritic arbor and the number of branch points were summed. The same measures from equivalent populations of neurons from the uninjured hemisphere and sham/naïve animals served as the control population. Approximately 90 neurons in each hemisphere per animal were digitized from the dentate gyrus, 20 neurons in each hemisphere per animal in the CA3 region, and 60 neurons in each hemisphere per animal in the CA1 region. In each control animal, neurons from both hemispheres were combined for approximately 150–180 neurons in the dentate gyrus, 40 neurons in the CA3 region, and 120 neurons in the CA1 region. The methods of data collection and analysis were identical for PND7 animals, with the focus only on the CA1 region.

Statistics

In PND17 animals, the average soma area, branch length, and branch points in neurons were compared. Sample size was based on preliminary data from the DG, CA3, and CA1, which indicated a group size of three to six animals were required to have 80 % power to detect a 10 % change from contralateral hemisphere with a statistical level of 0.05. Averages in PND17 naïve and sham animals were combined to create a single control group (n = 4). Each quantitative measure was subjected to statistical analysis by one-way ANOVA (GraphPad Prism 6). Neuron averages in each region of each animal were averaged for statistical analysis. Planned post hoc comparisons compared groups to identify significant differences between groups at the p < 0.05 level.

In PND7 animals, where six animals were injured and two were shams, statistical analysis by t test was only performed on the six injured animals, comparing ipsilateral to contralateral measurements. Data from sham animals are included in the graphs, but not the statistical analysis. No statistical comparisons were conducted between PND17 and PND7 data.

Results

Soma area increased in the dentate gyrus 28 days following CCI injury

The soma area, cumulative dendritic length, and cumulative branch points of neurons in the DG were compared to the contralateral hemisphere and control animals in order to assess the impact of a severe CCI injury on neuron morphology at PID28 (Fig. 1). Soma area significantly increased in the ipsilateral hemisphere by 7.8 % in comparison to control animals (F(2,9) = 7.183; p = 0.01; Fig. 1c). The average soma area was 197.2 ± 2.3 μm2 in controls, 203.7 ± 0.6 μm2 in the contralateral hemisphere, and 212.6 ± 4.4 μm2 in the ipsilateral hemisphere. No significant changes in morphology were detected in dendritic length (F(2,9) = 2.293; p = 0.16; Fig. 1d) or number of branch points (F(2,9) = 3.622; p = 0.07; Fig. 1e). The average dendritic length ranged from 758 to 1,119 μm and average branch points ranged from 4.8 to 6.5 per neuron across all groups. These data indicate that soma area increased slightly in the ipsilateral DG, without change in dendritic architecture at 28 days after CCI compared to uninjured controls.

Fig. 1
figure 1

For the dentate gyrus of the hippocampus, soma area significantly increased in the ipsilateral hemisphere of PND17 rats at 28 days after CCI compared to control (C). (A) Bright field images of Golgi stained dentate gyrus neurons from control (A1), contralateral hemisphere (A2) and ipsilateral hemisphere (A3) at PID28. (B) Representative neuron reconstructions from each location traced with the Neurolucida program. (D, E) No significant differences were measured in dendritic length or branching points between groups. Controls are the average of uninjured sham (n = 2) and naïve (n = 2) animals. Data were analyzed by one-way ANOVA with Tukey’s post hoc comparison. Significance, *p < 0.05 compared to controls. Bar graphs represent the mean ± SEM with four animals per group. Magnification ×20

Full size image

It is important to note that our preliminary observations indicated differences between the ventral and dorsal blade of the DG, with the dorsal blade more labile to morphological change after injury as measured by larger soma and shorter dendritic arborization [34]. This distinction was not borne out in the complete dataset; therefore, DG neurons from the ventral and dorsal blade have been pooled for analysis.

Soma area increased in the CA3 at 28 days following CCI injury

Morphological parameters of neurons in the CA3 were compared to the contralateral hemisphere and control animals in order to assess the impact of a severe CCI injury in PND17 animals on neuron morphology in the CA3 at 28 days post-injury (Fig. 2). Soma area significantly increased in the ipsilateral hemisphere by 11.3 % in comparison to control animals (F(2,9) = 4.657; p = 0.04; Fig. 2c). The average soma area was 401.8 ± 5.5 μm2 in controls, 414.7 ± 15.6 μm2 in the contralateral hemisphere, and 447.2 ± 8.9 μm2 in the ipsilateral hemisphere. No significant changes in morphology were detected in dendritic length (F(2,9) = 0.225; p = 0.80; Fig. 2d) or number of branch points (F(2,9) = 0.7594; p = 0.4957; Fig. 2e). The average dendritic length ranged from 1,693 to 3,088 μm and average branch points ranged from 13.6 to 24.1 per neuron across all groups. As in the DG, these data indicate that severe CCI injury resulted in an increased soma area in the ipsilateral CA3 at 28 days post-injury compared to uninjured controls.

Fig. 2
figure 2

For the CA3 of the hippocampus, soma area significantly increased in the ipsilateral hemisphere of PND17 rats at 28 days after CCI compared to control (C). (A) Bright field images of Golgi stained dentate gyrus neurons from control (A1), contralateral hemisphere (A2) and ipsilateral hemisphere (A3) at PID28. (B) Representative neuron reconstructions from each location traced with the Neurolucida program. (D, E) No significant differences were measured in dendritic length or branching points between groups. Controls (N = 4) are represented by uninjured sham (n = 2) and naïve (n = 2) animals. Data were analyzed by one-way ANOVA with Tukey’s post-hoc comparison. Significance, *p < 0.05 compared to controls. Bar graphs represent the mean ± SEM of four animals per group. Magnification ×20

Full size image

Cumulative dendritic length and branching points decreased in the CA1 28 days following CCI injury

Morphological parameters of neurons in the ipsilateral brain-injured CA1 were compared to the contralateral hemisphere and control animals in order to assess the impact of a CCI injury in PND17 animals at 28 days post-injury (Fig. 3). No significant changes in soma area were detected (F(2,9) = 0.58; p = 0.58; Fig. 3c). The average soma area ranged from 287.2 to 301.2 μm2 per neuron across all groups.

Fig. 3
figure 3

For the CA1 of the hippocampus, dendritic length (D) and branching points (E) were significantly reduced in the ipsilateral hemisphere of PND17 rats at 28 days after CCI compared to control and the contralateral hemisphere. Significantly fewer branching points were also measured in the contralateral hemisphere compared to control (E). (A) Bright field images of Golgi stained CA1 neurons from control (A1), contralateral hemisphere (A2) and ipsilateral hemisphere (A3) at PID28. (B) Representative neuron reconstructions from each location traced with the Neurolucida program. (C) Soma area was not significantly different between groups. Data were analyzed by one-way ANOVA with Tukey’s post hoc comparison. Significance, *p < 0.05 compared to controls and +p < 0.05 compared to contralateral hemisphere. Bar graphs represent the mean ± SEM of four animals per group. Magnification ×20

Full size image

Dendritic length significantly decreased in the ipsilateral hemisphere by 20–26 % in comparison to controls and the contralateral hemisphere, respectively (F(2,9) = 13.36; p = 0.002; Fig. 3d). The average dendritic length was 2,564 ± 85.9 μm in controls, 2,369 ± 63.0 μm in the contralateral hemisphere, and 1,888 ± 125.7 μm in the ipsilateral hemisphere.

The number of branching points significantly decreased in the ipsilateral hemisphere by 26.7 % in comparison to control animals and by 15.5 % in comparison to the contralateral hemisphere (F(2,9) = 32.98; p < 0.0001; Fig. 3e). In the contralateral hemisphere, there was a 13.3 % reduction in the number of branch points in comparison to control animals. The average cumulative branching points were 22.6 ± 0.5 in controls, 19.6 ± 0.4 in the contralateral hemisphere, and 16.6 ± 0.5 in the ipsilateral hemisphere.

These data indicate that dendritic length and number of branching points in the ipsilateral hemisphere of the CA1 are susceptible to severe CCI at 28 days post-injury, without change in soma area.

No significant morphological changes in CA1 region of PND7 animals

The soma area, cumulative dendritic length, and cumulative branch points of neurons in the CA1 of PND7 animals were compared to the contralateral hemisphere in order to assess the impact of a severe CCI injury on morphology in the CA1 at 28 days post-injury (Fig. 4). No significant changes in morphology were detected in soma area between the ipsilateral and contralateral hemispheres (F(2,11) = 1.406; p = 0.29; Fig. 4a), with an average area ranging from 310.1 to 311.4 μm2 per neuron across both groups (Fig. 4). No significant changes were detected for cumulative dendritic length (F(2,11) = 1.44; p = 0.28; Fig. 4b) or cumulative branching points (F(2,11) = 2.03; p = 0.18; Fig. 4c). The average cumulative dendritic length and branch points ranged from 2,195 to 2,546 μm per neuron and 19.3 to 22.8 per neuron across both groups, respectively. Calculations for sham animals were included in the figure, but not used for statistical analysis.

Fig. 4
figure 4

CCI in PND7 rats did not cause significant change to the (a) soma area, (b) dendritic length, and (c) branching points at 28 days post-injury in the CA1 region of the hippocampus in comparison to the contralateral hemisphere. Sham (n = 2); contralateral (n = 6); ipsilateral (n = 6). These data were analyzed with a one-way ANOVA with Tukey’s post-hoc comparison. Bar graphs represent the mean ± SEM. Magnification ×20

Full size image

Discussion

In this study, we have shown that experimental TBI using CCI on PND17 rats results in increased soma area in the DG (7.8 %) and CA3 (11.3 %) regions of the hippocampus at PID28 in comparison to controls. Additionally, there was a 20–26 % injury-induced reduction in dendritic length and dendritic branching in the CA1 region in comparison to both the contralateral hemisphere and to controls. In younger PND7 rats, none of these morphological changes were observed in CA1, indicating an age-at-injury dependence on this morphological response to injury. These data indicate long-lasting morphological alterations of neurons in the hippocampus resulting from injury at PND17 that could impact continued brain development and influence the progression of post-injury cognitive function.

The increase in cell body size in the DG and CA3 may be interpreted as neuronal hypertrophy or selective loss of small neurons in these regions [35]. Hypertrophy is often considered a pathological event; however, insufficient evidence exists to correlate soma size to disease state [36]. Neuron hypertrophy can also result from increased metabolic or transcriptional demands, perhaps as compensation for neuron cell loss or accommodation of more neurons within the hippocampal trisynaptic pathway and subsequent synaptogenesis, as we previously reported in PND17 rats at PID28 [25, 35]. The absence of change in dendritic length and branch points in these regions is antithetical to the correlation between soma size and extent of dendritic arborization [35]. However, the neuronal and synaptic response to injury concomitant with developmental processes could negate this relationship. Assessment of neuron pathology and morphology at later-time points may reveal a role for hypertrophied soma in hippocampal circuit function in contrast to CA1.

To date, Nissl stains have been the principal means to define neuronal loss and contusion volume in experimental TBI. Yet, the Nissl selectively stains the endoplasmic reticulum and Nissl substance of the perikarya, while leaving neuronal processes essentially unstained. The Golgi stain can be used to visualize and characterize the morphology of the dendritic arbors, whose structure (branch length and branch points) directly reflects the magnitude of afferent input. The fine discrimination of individual neurons may better represent trauma-induced alterations in neuronal morphology, particularly as it relates to circuit function and behavioral performance.

Mechanical forces of injury induce differential strains on the CA1 and CA3 during CCI, which may explain the regional differences in morphology. Tensile strains are applied over a larger area and to a greater magnitude in the CA3 compared to CA1, indicating that CA3 may be more vulnerable to injury-inducing forces despite CA3 lying farther from the site of impact [37]. Additionally, significant CA3 neuronal loss has been repeatedly demonstrated after CCI [38–42], yet the extent of dendritic damage to either region has not been addressed, particularly in the juvenile animal. With the proximity of CA1 to the site of impact and the reduced incidence of cell death, it is plausible that CA1 neurons may preferentially sustain dendritic damage without cell death. Moreover, cell loss from the CA3 would result in afferent loss in the CA1, thereby influencing the extent of CA1 dendritic arborization [41].

Previously, we studied circuit reorganization in the hippocampus at PID28 following CCI in PND17 rats using viral transneuronal tracing [25]. This study demonstrated that more neurons were incorporated into hippocampal circuitry of the injured hemisphere compared to that of sham-operated controls. Injury-induced sprouting may expand the terminal arbors of neurons, ultimately incorporating more neurons into topographically organized circuits. Alternatively, neurons culled from the circuit by cell death could be compensated for in the normal course of development. In either condition, it is important to note that cognitive performance of injured animals in the Morris Water Maze returned to sham levels at the time that the synaptology of the circuit was analyzed (PID28) despite CCI-injured PND7 and PND17 rats displaying significant cognitive deficits at PID11–17 in comparison to sham rats [21, 25]. In diffuse TBI by impact acceleration in the juvenile, the extent of hippocampal reorganization, particularly as it relates to enduring cognitive deficits, is unknown [26, 27]. Thus, the injured developing brain may have the capacity to reorganize in a manner consistent with functional recovery.

Multiple laboratories have demonstrated that early cognitive deficits following CCI return to sham levels by PID30 and remain through PID60 [21, 24]. Therefore, the observed circuit reorganization in terms of neuronal hypertrophy (DG, CA3) and dendritic pruning (CA1) may be indicative of an adaptive or reparative process, which facilitates long-term cognitive function. However, long-term performance on sensorimotor, balance, coordination, and turn bias tasks remain significantly impaired through PID60 [24], suggesting that the juvenile brain cannot compensate for all neuropathology resulting from brain injury. In contrast, severe diffuse brain injury in the juvenile induced lasting cognitive deficits [26] and acute pathology that resolved rapidly [27]. Late-onset and long-lasting functional deficits may result from maladaptive or inappropriate circuit reorganization.

No definitive studies can explain the differential responses in CCI-injured PND7 compared to PND17 rats; however, trauma-induced apoptotic neurodegeneration has been found to be enhanced in the DG of PND3 and PND7 rats in comparison to PND10, PND14, and PND30 after the weight drop model of diffuse brain injury [28]. Bittigau et al. suggest that the degree of myelination, brain water content, N-methyl-D-aspartate (NMDA) receptor hypersensitivity period in PND3 and PND7 neonates may leave them more vulnerable to mechanical and excitotoxic injury [43, 44]. Progressive axonal injury and secondary post-injury sequelae may also promote neuron apoptosis in PND3 and PND7 rats due to a combination of excitotoxic neurochemical exposure, circulation of pre-apoptotic molecules and a loss of trophic support due to a loss of afferents [45]. Simply, the younger animals can endure cell death associated with normal development, and the additional pathology from brain injury does not saturate these undefined coping mechanisms. Alternatively, neurons injured directly at PND7 may die, without consequence to surviving neurons, whereas neurons injured at PND17 survive but suffer altered morphology.

Summary

Quantifiable morphological changes occurred in PND17 and PND7 animals at PID28 following CCI, including increased soma area in the DG and CA3 and decreased dendritic length and branching points in CA1. The most robust response was found in CA1, but did not occur in animals injured at PND7, indicating an age-at-injury response. This study impacts pediatric neurotrauma research by demonstrating that there are both age-dependent and region-dependent morphological changes that result from CCI in juvenile rats. A clear understanding of the chronic structural consequences of brain injury can more completely inform and direct pediatric neurotrauma clinicians and researchers in their approach to improve care, treatment, and rehabilitation.