Traumatic brain injury modifies synaptic plasticity in newly-generated granule cells of the adult hippocampus

Abstract

The hippocampus is vulnerable to traumatic brain injury (TBI), and hippocampal damage is associated with cognitive deficits that are often the hallmark of TBI. Recent studies have found that TBI induces enhanced neurogenesis in the dentate gyrus (DG) of the hippocampus, and this cellular response is related to innate cognitive recovery. However, cellular mechanisms of the role of DG neurogenesis in post-TBI recovery remain unclear. This study investigated changes in long-term potentiation (LTP) within the DG in relation to TBI-induced neurogenesis. Adult male rats received a moderate TBI or sham injury and were sacrificed for brain slice recordings at 30 or 60 days post-injury. Recordings were taken from the medial perforant path input to DG granule cells in the presence or absence of the GABAergic antagonist picrotoxin, reflecting activity of either all DG granule cells or predominately newborn granule cells, respectively. Measurements of LTP observed in the total granule cell population (with picrotoxin) showed a prolonged impairment which worsened between 30 and 60 days post-TBI. Under conditions which predominantly reflected the LTP elicited in newly born granule cells (no picrotoxin), a strikingly different pattern of post-TBI changes was observed, with a time-dependent cycle of functional impairment and recovery. At 30 days after injury this cell population showed little or no LTP, but by 60 days the capacity for LTP of the newly born granule cells was no different from that of sham controls. The time-frame of LTP improvements in the newborn cell population, comparable to that of behavioral recovery reported previously, suggests the unique functional properties of newborn granule cells enable them to contribute to restorative change following brain injury.

Introduction

An estimated 1.1% of the U.S. population is living with a long-term disability resulting from a moderate to severe traumatic brain injury (TBI) (Zaloshnja et al., 2008). Cognitive deficits including learning and memory impairment are common among TBI patients that may persist for decades following injury (Wilson et al., 2017). Although many promising neuroprotective treatment options have been identified in preclinical studies, these primarily target the lesion size, and translation of effective experimental treatments to clinical trials has proven challenging. In a summary review covering 100 randomized and controlled clinical trials in adult TBI, acute phase pharmacologic or non-phamacologic trials showed no positive benefits on TBI outcome (Lu et al., 2012). While research into effective neuroprotective strategies is essential, there is growing interest in cell-based and pharmacological therapies that enhance endogenous neurorestorative processes by facilitating neurogenesis, synaptogenesis, angiogenesis, and axonal remodeling (e.g., review by Xiong et al., 2015). Work in our laboratory, and others, has demonstrated hippocampal neurogenesis to be an endogenous process which may be enhanced to benefit innate restorative processes after TBI (Dash et al., 2001; Richardson et al., 2007; Sun et al., 2007, 2009, 2015; Weston and Sun, 2018).

The dentate gyrus (DG) of the hippocampus is known to serve an essential role in learning and memory, while being extremely vulnerable to TBI. Studies that quantify neuronal degeneration following experimental TBI, routinely identify the DG as exhibiting especially high numbers of dead or dying cells (e.g., Anderson et al., 2005; Gao et al., 2008). In the DG, the subgranular layer (SGZ) is the site of hippocampal neurogenesis, where neural stem cells are continuously generated throughout life, with the majority differentiating into granule cell (GC) neurons and maturing to integrate into the existing hippocampal neural network (Ge et al., 2008; Imayoshi et al., 2008; Kempermann et al., 2015; Toni et al., 2015). Following TBI, the immature new neurons in the SGZ are vulnerable to injury showing prominent degeneration at 2 days post-injury (Gao et al., 2008) and likely undergoing apoptosis (Sun et al., 2013).

Although the immature GCs only comprise about 10% of the total GC number (Imayoshi et al., 2008) they exhibit a set of functional properties distinct from those of mature GCs, which enable the newly-generated cell population to exert a significant influence on DG functioning. These include an enhanced capacity for plasticity and a low threshold for LTP induction (Lemaire et al., 2012; Schmidt-Hieber et al., 2004), a low level of GABAergic inhibition (Wang et al., 2000), and a relatively high expression level of NR2B-containing NMDA receptors (Snyder et al. 2001; Ge et al. 2007). The heightened intrinsic excitability of the young cells also enables them to modulate the low activity level (sparse coding) of the mature GCs (Ikrar et al., 2013). Moreover, multiple learning tasks have been shown to specifically require intact hippocampal neurogenesis including contextual fear memory (Imayoshi et al., 2008), object recognition (Suarez-Pereira et al., 2014), and active place avoidance (Burghardt et al., 2012).

Our laboratory has investigated the effects of TBI on hippocampal neurogenesis, and explored the efficacy of intervening in this endogenous process as a neurorestorative strategy. We observed that following a lateral fluid percussion injury (LFPI) in adult rats, an experimental model of TBI, the ongoing generation of adult-born GCs was significantly increased over a two-week post-TBI period (Sun et al., 2005). An additional study examined the time required for the newly-generated GCs, increased in numbers by the LFPI, to form synapses on their postsynaptic target neurons in CA3 (Sun et al.,2007). This was an important step towards integrating the newborn cells into the hippocampal network, and the timing of this synaptogenesis closely matched the spontaneous recovery of a TBI-induced deficit in a spatial memory task using the Morris Water Maze (Sun et al.,2007). Blocking hippocampal neurogenesis with an antimitotic agent following the LFPI prevented the cognitive recovery of the maze performance at 56–60 days post-injury (dpi) (Sun et al., 2015), which strongly suggested a link between TBI-induced cell proliferation and cognitive recovery.

Although we have observed post-TBI cognitive recovery to be significantly dependent on neurogenesis, questions remain concerning the cellular basis of this association. To investigate functional changes in the newborn cell population, in this study we have assessed the capacity for long-term potentiation (LTP) in the DG following an LFPI in adult rats. LTP in the newly generated GCs was measured separately from that of existing mature GCs, enabled by the differential response to GABA antagonists in these two cell populations. It is well-established that the immature neurons do not exhibit the high baseline GABAergic inhibition observed in mature GCs, and functional properties of the new neurons are not significantly altered by GABA antagonists. Accordingly, multiple laboratories have obtained LTP recordings in the presence vs. absence of picrotoxin (Snyder et al., 2001) or bicuculline (Wang et al., 2000; Ge et al., 2008), with results indicating that all granule cells (mature and immature) are recorded in the presence of GABA blockers, but without the blockade evoked signals are generated predominantly by the immature neuronal population. The present study used picrotoxin to obtain LTP recordings separately for the immature vs. mature GCs at 30 and 60 days post-TBI, to provide comparisons with our previous behavioral findings. We report here a time-dependent restoration in the capacity for LTP in the population of adult-generated GCs after TBI, in contrast to a progressive loss of LTP capacity in the mature neuron population. These findings highlight the functional significance of new neurons generated following brain injury, and point to potential new therapeutic strategies in facilitating cognitive recovery.

Materials and Methods

Experimental Animals.

A total of 57 male 3-month old Sprague-Dawley rats (ENVIGO) weighing about 300g were included in these experiments. Animals were housed as pairs in an animal facility with controlled temperature and humidity, a 12 hour light-dark cycle, and food and water were provided ad libitum. All procedures were approved by the VCU Institutional Animal Care and Use Committee.

Brain slice recording was conducted in a total of 37 rats, which underwent TBI or a sham injury and were allowed to survive for either 30 or 60 days prior to recording. Rats were randomly assigned to recording conducted in the presence of picrotoxin [(+)Pic condition] or with no picrotoxin present [(−)Pic condition]. Group sizes for the (+)Pic condition were TBI-30 d (n=7), TBI-60 d (n=7), Sham-30d (n=3), Sham-60d (n=4). Group sizes for the (−)Pic condition were TBI-30 d (n=5), TBI-60 d (n=6), Sham-30d (n=2), Sham-60d (n=3). The sham rats recorded at 30 d and 60 d did not differ on any of the electrophysiological parameters measured, and were pooled for the statistical analysis for each condition.

Histological studies were conducted in a total of 20 rats, which underwent TBI or a sham surgery and survived for either 30 or 60 days. Group sizes were TBI-30 d (n=7), TBI-60 d (n=5), Sham-30 d (n=4), Sham-60d (n=4). The sham rats sacrificed at 30 d and 60 d did not significantly differ on any of the anatomical regions measured, and were pooled for the statistical analysis.

Surgical Procedure.

Animals received a moderate lateral fluid percussion injury (LFPI) or sham surgery following our published protocol (Sun et al., 2007, 2015). Briefly, the rats were placed in an acrylic induction chamber to anesthetize with 4% isoflurane. To stabilize their heads in preparation for surgery, the rats were placed in a ventilated stereotaxic frame with an inlet containing 2.5% isoflurane in a gas mixture (30% O₂ and 70% N₂). A midline incision was made to expose the skull, and a 4.9mm trephine was used to make a craniectomy on the left parietal bone halfway between the lambda and bregma sutures. A Luer-slip style hub was made from a 20-gauge needle cap and placed to the craniectomy site with cyanoacrylate and reinforced with dental acrylic to the skull, then anesthesia was turned off. As animals regained consciousness the Luer-slip style hub was filled with saline and connected to a pre-calibrated FPI device and a pulse of 2.01 ± 0.05 ATM was administered to induce a moderate TBI. Immediately after injury the hub was removed and animals were returned to the surgical table to monitor righting time. Sham control rats received the same surgical procedure but without the fluid pulse. After injury or sham surgery the incision site was sutured and the animals were returned to their cage.

Electrophysiological Recording Procedures.

At 30 or 60 days following FPI or sham injury, rats were anesthetized with 4% isoflurane for 4 minutes and the brain rapidly removed. Horizontal 400-μm slices were taken from the hemisphere ipsilateral to injury, and cut into low-calcium artificial cerebrospinal fluid (ACSF: 2–4 °C) containing (in mM) NaCl 124, KCl 3,MgSO4 5, NaHCO3 26, NaH2PO4 1.25, glucose 10, saturated with 95% O2/5% CO2. Slices were then transferred to an equilibration chamber containing oxygenated ACSF used for electrophysiological recording containing (in mM) NaCl 124, KCl 3, MgSO4 2, NaHCO3 26, NaH2P4 1.25, glucose 10, CaCl2 2, and either 50 μM picrotoxin or no picrotoxin. Slices were allowed to equilibrate for at least one hour prior to recording. For recording, one slice per brain was recorded. Briefly, the slice was transferred to a submersion-type recording chamber perfused at a rate of 2 ml/min with the solution used for equilibration at 32°C. A bipolar stimulating electrode and a micropipette recording electrode (filled with ACSF, resistance 2–4 MΩ) were positioned in the DG molecular layer to evoke field excitatory postsynaptic potentials (fEPSPs) through the medial perforant pathway (MPP). To verify placement in the MPP, and not the lateral perforant pathway, responses to paired stimulus pulses (interpulse intervals of 50, 100 and 150 ms) were confirmed to produce paired-pulse depression or no change (Wang et al., 2016). Paired-pulse depression or facilitation, when evoked with the low current levels used here (sub-threshold for population spikes), reflect mainly presynaptic short term plasticity (Lomo, 1971; Zucker, 1989). Our analyses confirmed that responses to paired stimulation were not systematically different among experimental groups, to provide further evidence that TBI-induced changes in LTP primarily reflected alterations at the postsynaptic level. For LTP experiments, a stimulating intensity that evoked fEPSPs of 50% maximum, as based on input-output testing, was delivered at a rate of one pulse every 30 s and used to obtain a 30-min period of baseline recording. LTP was then induced using three trains of theta-burst high frequency stimulation (HFS), consisting of 10 bursts of four pulses at 100 Hz, with 200 ms separating the onset of each burst. Each train was separated by 50 s. Following HFS, fEPSPs were acquired for 60 min using stimulus parameters identical to those of the baseline recording. For LTP baseline and post-HFS data, mean fEPSP slopes were aggregated into 2-min epochs for graphical and statistical analyses.

Brain tissue histological assessment.

Following TBI, injured brain often undergoes atrophy. To assess the degree of this change at 30 or 60 days after injury, a separate cohort of rats were anesthetized and transcardially perfused with phosphate buffer saline (PBS) followed by 4% paraformaldehyde. Brains were harvested and post-fixed in 4% paraformaldehyde and stored at 4° C in PBS+0.05% sodium azide until processed for experiments. Coronal 60μm slices were collected in 24-well plates containing PBS+0.05% sodium azide and stored at 4° C until staining. For visualization of tissue structure in volume ratio measurements, slices were processed for hematoxylin and eosin (H&E) staining (Smith et al., 1997). Every 6th slice in sequence beginning near −2.30 mm from Bregma were taken until a total of 6 slices per animal were collected and mounted on slides, providing a thorough representation of both the cortex and hippocampus. Images were taken with a widefield microscope to visualize the entire slice. Tissue volume of the dentate gyrus, total hippocampus, and cortex were measured using Image J program.

Briefly, Image files of H&E stained slices were imported into FIJI software for analysis using the plugin TrakEM2 (Cardona et al., 2012). Files for all 6 slices per animal were grouped together as a sequence and correct calibration values were set based on pixels per μm. Within each layer the z-resolution was adjusted based on magnification of the microscope camera, original slice thickness in μm, and distance apart each layer was respective of each other within the Z coordinate. The software generated an image sequence for each animal where we could selectively highlight a specific region of interest. Values from the highlighted regions were generated to create a total area measurement for ipsilateral and contralateral hemispheres of the dentate gyrus, total hippocampus, and cortex.

Statistical analysis.

The effects of injury condition (sham-injury, TBI 30 dpi, TBI 60 dpi) on LTP magnitude and evoked paired-pulse responses were evaluated using ANOVA (SPSS v21 – GLM Procedures) combined with Bonferroni post hoc tests. The significance of shifts in input-output curves was evaluated with the Mann-Whitney U test. A nonparametric approach was selected in the latter case, because homogeneity of variance requirements were not met when ranges of stimulus intensities were applied. The % IPSI / CONTRA representing brain tissue atrophy (sham-injury, TBI 30 dpi, TBI 60 dpi) was evaluated using ANOVA (JMP Pro v15) combined with Tukey HSD post hoc tests. The significance level, α = 0.05, was used for all inferential statistics, and averaged values are expressed as mean ± SEM.

Results

The Capacity for LTP Decreases over Time in the Presence of Picrotoxin

There is extensive evidence that GC excitation is strongly modulated by GABAergic inhibition, and bath application of GABA antagonists have been routinely used by multiple laboratories to facilitate LTP induction in the perforant path input to the DG (e.g., reviewed by Bortolotto et al., 2011). We examined the increase in LTP magnitude caused by the bath application of picrotoxin in our sham-injured groups (a control condition which does not encroach on brain tissue, and is highly similar to naïve control animals). Similar to numerous prior reports based on uninjured naïve animals, the LTP magnitude induced in our (+)Pic sham rats (246% ± 26.9% of baseline levels) was strikingly elevated over that of the (−)Pic sham group (131% ± 14.0% of baseline) [F(1,10)=11.243; p<0.01]. This disparity represents an 88% increase in LTP magnitude due to the presence of picrotoxin. The effect of the GABAergic blockade was especially evident in the response increases induced by the high frequency condition trains, but was less potent in affecting the baseline (pre-HFS) amplitude of the evoked signal. Baseline fEPSP mean slope values for (+)Pic sham rats was 0.47 ± 0.10 mV/ms, which was 64% higher than (−)Pic mean fEPSP slopes (0.29 ± 0.08 mV/ms), but this difference was not significant (F(1,10)=1.94; n.s.).

We measured LTP in the MPP at 30 dpi and 60 dpi to provide useful comparisons to our prior behavioral findings that cognitive deficits remained present at 30 dpi, and spontaneous cognitive recovery was observed at 60 dpi (Sun et al., 2007). Mean fEPSP slope values (expressed as percent of pre-HFS baseline) are plotted in Figure 1 for the (+)Pic condition, showing the final 10 minutes of baseline recording and the 60 minutes of post-conditioning recording for sham-injured, and TBI rats measured at 30 dpi and 60 dpi. ANOVA analyses showed overall significant differences among the analytic groups, with the omnibus F(2,18)=3.89; p<0.05. The theta-burst HFS induced a robust magnitude of LTP in the sham-injury cases, with a mean potentiation of 246% ± 26.9% of baseline, averaged across the 60 min. post-HFS recording period. The effects of TBI were observed as a time-dependent decrease in the capacity for LTP. At 30 dpi, the mean level of LTP was 179% ± 24.1% of baseline, and at 60 dpi the magnitude of LTP had further decreased to a mean of 145% ± 27.2% of baseline. Bonferroni-corrected pairwise comparisons showed the 60 dpi level of LTP to be significantly below the sham-injury level (p<0.05); the 30 dpi level of potentiation did not differ significantly from either of the other groups. Because stimulus pulses applied to the disinhibited (+)Pic slices recruit the full complement of mature and immature GCs, these results likely reflect a TBI-induced loss of function, affecting most GCs, and which may continue to worsen even through 60 days following the injury.

Figure 1. LTP measured in the presence of picrotoxin (+)Pic.

Mean fEPSP slopes (+SEM), expressed as percent of baseline level, are plotted for sham-injury, TBI 30 dpi, and TBI 60 dpi, showing the final 10 minutes of baseline recording and 60 minutes of post-HFS recording. Responses were evoked and collected at a rate of 1/30 sec, but for statistical and graphical analyses were aggregated into 2-minute epochs. The magnitude of LTP in the sham injury group, averaged over the 60 minutes of post-HFS recording, was significantly larger than in the TBI 60 dpi group (*p<0.05

This “time-series” approach to monitoring LTP levels, stimulating with a constant current level throughout the full baseline and post-HFS phases, provides a sensitive index of response stability over time. However, it is also useful to consider evoked responses acquired during input-output testing. As stimulus pulses are applied in a graduated series of increasing current strength, the effective radius of stimulation is progressively increased, with successive pulses recruiting a growing number of cells into the field potential. We obtained input-output series at the beginning of baseline recording to determine the half-amplitude standard current for HFS, and again at 60 min following HFS. Comparing these curves shows how LTP is expressed over a stimulus range. Figure 2A shows representative waveforms for the (+)Pic groups, and corresponding input-output curves (Figure 2B). HFS induced significant leftward shifts of the input-output curves for sham and 30 dpi, but not for slices at 60 dpi, which closely matched the pattern of LTP magnitudes seen in the time-series of Figure 1.

Figure 2. Analysis of Input-Output relationships in (+)Pic study groups.

A. Evoked waveforms show representative input-output series evoked in sham and injury groups during baseline recording and at 60 min post-LTP. B. Mean fEPSP slopes (+SEM) are plotted for graded current intensities, spanning threshold to the maximum response obtained during baseline recording. Significant tetanus-induced upward shifts were observed for sham and TBI-30d (+)Pic slices (Mann-Whitney U statistic shown as Z values).

Newly Born GCs Increase their Capacity for LTP following TBI

We found that recordings in the absence of picrotoxin [(−)Pic condition], produced a smaller magnitude of LTP in our sham-injured rats relative to (+)Pic sham-injured cases, which is consistent with prior findings (Snyder et al., 2001). In addition, this dependence of LTP on the status of GABAergic inhibition, is consistent with the accumulated evidence that the newly born GCs are insensitive to GABAergic inhibition and readily express functional plasticity. Baseline and post-HFS mean fEPSP slopes are plotted for each (−)Pic analytic group in Figure 3. ANOVA analyses showed overall significant differences in LTP among the groups (sham, 30 dpi, 60 dpi) [F(2,13)=3.97; p<0.05]. Recordings in sham-injured (−)Pic rats showed HFS induced a level of LTP of 131% ± 13.5% relative to pre-HFS baseline. Interestingly, TBI led to a markedly different pattern of change over time when compared to the (+)Pic condition. At 30 dpi, essentially no LTP could be induced in (−)Pic slices, with the post-HFS amplitudes being only 103% ± 6.6% of baseline. At 60 dpi, the level of LTP observed was 150% ± 10.7%, a value higher even than the sham-injury level. However, post-hoc pairwise comparisons showed that the mean level of LTP at 60 dpi was significantly different only from the 30 dpi level (p<0.05), and the magnitude of LTP seen in the sham-lesion group was not significantly different from either of the injury groups (Figure 3). Example input-output waveforms and associated curves are shown in Figure 4, although this index of potentiation proved less sensitive to detect LTP at the more modest level of plasticity observed in the (−)Pic condition. The dramatic recovery of the capacity for LTP, observed to occur between 30 dpi and 60 dpi, may well represent an important resource in the search for therapies to facilitate cognitive recovery from brain injury.

Figure 3. LTP measured in the absence of picrotoxin (−)Pic.

Mean fEPSP slopes (+SEM), expressed as percent of baseline level, are plotted for sham-injury, TBI 30 dpi, and TBI 60 dpi, showing the final 10 minutes of baseline recording and 60 minutes of post-HFS recording. Responses were evoked and collected at a rate of 1/30 sec, but for statistical and graphical analyses were aggregated into 2-minute epochs. The magnitude of LTP in the TBI 60 dpi group, averaged over the 60 minutes of post-HFS recording, was significantly larger than in the TBI 30 dpi group (*p<0.05).

Figure 4. Analysis of Input-Output relationships in (−)Pic study groups.

A. Evoked waveforms show representative input-output series evoked in sham and injury groups during baseline recording and at 60 min post-LTP. B. Mean fEPSP slopes (+SEM) are plotted for graded current intensities, spanning threshold to the maximum response obtained during baseline recording. For the lower signal amplitudes of the (−)Pic group, the tetanus-induced curve shifts were not significant (Mann-Whitney U statistic shown as Z values).

In addition to the analyses of injury condition and survival interval on LTP plasticity, it was essential to establish there were no differences among the groups in basal excitability, which may have existed prior to the HFS conditioning trains. One indication there were no systematic differences in this factor was the lack of differences in fEPSP amplitude evoked with a standard stimulus current (100 μA) observed during input-output testing. Although input-output stimulus ranges were determined individually for each slice (based on the maximum evoked response), most of the rats (33 out of the total 37) included 100 μA in this range. There no significant differences in the mean fEPSP slopes for the (+)Pic groups (sham injury, 30 dpi, 60 dpi) [F(2,16)=1.314; p=0.296] or the (−)Pic groups [F(2,11)=1.517; p=0.262] in response to 100 μA stimulation (Figure 5). A second indication, that LTP differences were not attributable to between-group basal excitability variability, was the similarity in mean current levels used to evoke the half-amplitude fEPSP responses during baseline and post-conditioning recording. During input-output testing, a specific current intensity was found individually for each slice that elicited a fEPSP slope of 50% of maximum response. We found there were no significant differences in the mean current levels to achieve this 50% response for the (+)Pic groups [F(2,18)=0.366; p=0.698], or for the (−)Pic groups [F(2,13)=0.103; p=0.903]. These similarities among groups, provided support for interpreting any observed changes in LTP as reflecting the major study variables of injury condition and survival interval, and were not due to differences in basal excitability.

Figure 5. Analysis of responses evoked using 100μA pulses.

A. Representative responses, from each of the (+)Pic and (−)Pic groups evoked with uniform 100μA pulses, acquired during input-output recording. B. Mean group fEPSP slopes were not significantly different for either (+)Pic or (−)Pic conditions, suggesting basal excitability did not differ among study groups.

An examination of paired-pulse responses was also a prerequisite to the LTP analysis, confirming our expected signaling through the MPP, and clarifying that LTP changes reflect the physiology of GCs (newborn and mature), rather than exclusively presynaptic mechanisms. First, the placement of stimulating electrodes into the brain slices were aimed to the MPP, which has been reported to produce a characteristic paired-pulse depression or no change to identical paired pulses, whereas the lateral perforant path shows facilitation (Christie and Abraham, 1994; Wang et al., 2016). Assessment of our paired-pulse data showed no significant facilitation, and there were no differences in the ratio [fEPSP(2) / fEPSP(1)] at the three interpulse intervals tested (50, 100, and 150 ms) among the (+)Pic groups [F(2,18) = 1.07, p=0.364)], or (−)Pic groups [F(2,13)=0.108; p=0.898] (Figure 6). These results are consistent with our stimulating electrode placement in the MPP. Secondly, at the low current amplitudes used in our paired-pulse trials (sub-threshold for eliciting a population spike in the field potential), paired-pulse depression or facilitation reflects primarily pre-synaptic mechanisms (Lomo, 1971; Zucker, 1989). Our finding of no between-group differences in paired-pulse responding, helps clarify that injury-induced changes in LTP magnitude likely reflect changes at the postsynaptic level, i.e., in the functioning of granule cells, and not systematic group differences in presynaptic mechanisms.

Figure 6. Analysis of responses to paired-stimulus presentations.

A. Paired responses in a representative (+)Pic sham rat, evoked at interpulse intervals of 50, 100, and 150 ms. B. Group mean paired-pulse responses, plotting the ratio [fEPSP (2)/fEPSP(1) × 100] at the three interpulse intervals. There were no differences among groups in this index of presynaptic excitability for either (+)Pic groups or (−)Pic groups.

TBI Induces Persistent Brain Tissue Atrophy

TBI is typically associated with brain tissue atrophy that progresses after the initial injury. In order to assess this trait in our animals at the time when LTP was recorded, we measured the changes of volume in DG, total hippocampal, and cortex by comparing these regions ipsilateral (IPSI) to the injury to the contralateral (CONTRA) hemisphere. ANOVA analyses showed overall significant differences in volume ratio (% IPSI / CONTRA) among the groups for cortical tissue [F(2,17)=5.36; p<0.05], total hippocampus tissue [F(2,17)=5.86; p<0.05], and DG tissue [F(2,17)=11.25; p<0.001] (Figure 7). For the cortex, the mean IPSI/CONTRA ratio at 60 dpi was 84% ± 6%, which was significantly lower than the sham control level at 98% ± 2% (p<0.05), the mean for 30 dpi was 88% ± 2% which was not significantly different from sham or 60 dpi. For the volume of total hippocampus, the mean IPSI/CONTRA ratio at 60 dpi was 78% ± 3% which was significantly lower than the sham control level (94% ± 3%, p<0.05) and 30 dpi (92% ± 4%, p<0.05), the mean for 30 dpi was not significantly different from sham. For DG, significant difference was found between injured groups to sham (the mean IPSI/CONTRA ratio for 60dpi at 81% ± 2%, for 30dpi at 86% ± 3%, and sham at 97% ± 3%; p<0.01 for both injury groups versus sham), no difference was found between 30- and 60dpi groups. The representative tissue atrophy in the ipsilateral hemisphere is shown in Figure 7A–C. These measurements of tissue atrophy are consistent with the decline in LTP magnitude observed in the total GC neuron population [(+)Pic], with TBI-induced deficits seen at 30 dpi, and then further impaired at 60 dpi.

Figure 7. Changes in tissue volume after LFPI.

A-C. Representative H&E stained sections showing sham injury (A), 30 dpi TBI (B), and 60 dpi TBI (C). A delayed tissue shrinkage was observed ipsilateral to the injury (arrowhead). Group means for the volume ratio (%IPSI / CONTRA) are plotted below for cortical tissue (D), hippocampal tissue (E), and dentate gyrus tissue (F). D: For cortical tissue, the volume ratios observed at 60 dpi were significantly below sham values (*p<0.05). E: For total hippocampus, the volume ratios observed at 60 dpi were significantly lower than the sham and 30 dpi (*p<0.05), whereas no difference was found between 30 dpi and sham. F: For DG, the volume ratios in injured groups at 30- and 60 dpi were significantly lower than the sham group (**p<0.01), no difference was found between 30- and 60 dpi. Bar = 2mm.

Discussion

This study examined time-dependent changes in the capacity for LTP in the perforant path input to dentate granule cells following an experimental TBI in adult rats. A primary objective was to better understand the role of hippocampal neurogenesis in recovery from TBI. Because LTP is a leading cellular model of learning and memory, and cognitive dysfunction is among the most prevalent and debilitating symptoms of TBI (Nicholl and LaFrance, 2009), it was of particular interest to assess LTP plasticity in newborn GCs generated after the experimental injury. A separate assessment of LTP in immature GCs was enabled by their delayed development of a hyperpolarizing response to GABA, and their insensitivity to GABAergic antagonists (Ge et al., 2006, 2008; Snyder et al, 2001; Toni et al., 2015; Wang et al., 2000). When recordings were conducted in the presence of picrotoxin, with evoked potentials reflecting the activity of both mature and immature GCs, we found a TBI-induced decline in the LTP magnitude at 30 dpi, which was further deteriorated at 60 dpi. A contrasting pattern of post-injury LTP changes were observed in the absence of picrotoxin, which likely represented only immature GC activity, where the capacity for LTP was lost at 30 dpi, but by 60 dpi a robust level of LTP was exhibited.

Measurements obtained from the control condition in the present study, sham-injured rats, provide a basis for comparison with prior findings from other laboratories. The sham-injury procedure disturbed only a localized area of scalp and underlying bone, and did not invade the brain tissue itself. Accordingly, the electrophysiological properties of the present sham-injury condition are comparable to those of control (untreated) animals reported by other laboratories. We found a large magnitude (246% of baseline) LTP could be induced in the presence of picrotoxin, but a mean LTP of only 131% of baseline was evoked in the absence of picrotoxin. This disparity corresponds well with reports from many laboratories that consistent and large magnitude LTP is difficult to obtain in the in vitro DG without the use of GABAergic blockers (Bortolotto et al., 2011; Hanse and Gustafsson, 1992; Nguyen and Kandel 1996). The present results are also consistent with prior studies using picrotoxin (Snyder et al., 2001) or bicuculline (Wang et al., 2000; Ge et al., 2008) to differentiate these two populations: mature GCs under strong basal GABA inhibition vs. disinhibited young cells.

Neurogenesis impacts recovery from TBI and LTP plasticity

Studies in our laboratory and others have demonstrated that TBI enhances neurogenesis in the rodent hippocampus following TBI (Dash et al., 2001; Gao et al 2009; Sun et al., 2005, 2007, 2009, 2015). These studies have all labeled dividing adult-born GCs with the thymidine analog, bromodeoxyuridine (BrdU), administered after an experimental TBI. Our group has previously found that following a moderate LFPI in rats, TBI induced a significant increase in cell proliferation in the DG, with the peak reached at 2 dpi and a return to normal levels by 2 weeks post injury (Sun et al, 2005). We found that the majority of injury-induced proliferative cells survived and differentiated into mature GCs when examined at 10 weeks following TBI (Sun et al., 2007). Additionally, many of these post-injury generated GC neurons formed synaptic connections (Sun et al., 2007). We also observed that the time course for recovery of spatial memory performance at 56–60 dpi closely matched the establishment of new synapses from young GC axons onto CA3 pyramidal cells (Sun et al., 2007). Inhibiting proliferation with an antimitotic agent (arabinofuranosyl cytidine) following the LFPI completely abolished the innate cognitive recovery on Morris Water Maze performance at 56–60 dpi (Sun et al., 2015), demonstrating a close association between TBI-induced cell proliferation and cognitive recovery. Further evidence, supporting a role for injury-induced neurogenesis in recovery of cognitive function, was obtained by administering basic fibroblast growth factor (bFGF), during the first post-injury week after the LFPI. The bFGF treatment enhanced neurogenesis in the DG, and significantly improved cognitive performance on days 22–25 following injury (Sun et al., 2009).

In contrast to TBI-induced increases in neurogenesis, there is also evidence that experimental TBI leads to cell death affecting specific granule cell types. After CCI in mice, a significant cell death of immature neurons (Doublecortin-positive) was observed at 4 hr – 72 hr, although degenerating cells were rarely observed in the subgranular zone (Gao et al., 2008). Another study investigating CCI in mice reported a postinjury loss of Doublecortin-positive immature neurons, but an injury-induced increase in Nestin-expressing early progenitor cells (Yu et al., 2008). Further, the activated early neural progenitors later repopulated the lost immature neuron numbers (Yu et al., 2008). Following a moderate LFPI in rat, we have found heighted level of injury-induced apoptosis in the GCs cells, and an age-related difference in cell type vulnerability with SGZ immature neurons (PSA-NCAM+) more vulnerable in juvenile, whereas GCs mature neurons (NeuN+) more vulnerable in the aged brain (Sun et al., 2013).

Inhibitory interneurons in the DG, particularly in the hilus region, are among the most vulnerable cell populations to TBI. The extended cytoarchitecture of a typical interneuron, with a dense axonal plexus and elaborate dendritic arbor, is a risk factor to the stretch and shear forces of concussive injuries. Studies using the same LFPI model have documented TBI-induced death of various types of DG inhibitory neurons. These include parvalbumin-expressing basket cells, with post-injury losses ranging from 38% to 67% (Toth et al., 1997; Huusko et al., 2015), and losses of somatostatin-immunoreactive cells, first noted in the hilus by Lowenstein et al. (1992), and later quantified as an approximate 54% reduction by Huusko et al. (2015). Interneurons which survived TBI often exhibit functional changes which, in turn, impact network and plasticity function. For example, surviving interneurons in the granule cell layer were reported to exhibit a depolarizing shift of about 10 mV, which lasted up to 4 days after LFPI (Ross and Soltesz, 2000). It is well established that postsynaptic sites, vacated by degenerating synaptic terminals, are then occupied by sprouting terminals from nearby presynaptic sources. Hunt and colleagues (2011) examined post-TBI sprouting-related reorganization in the DG after a CCI injury in mouse, which culminated in increased excitatory drive to surviving hilar GABAergic neurons. One consequence of such rewiring is potential network destabilization with aberrrant interneuronal control of granule cell excitability, and resulting abnormalities in neuroplasticity function including LTP.

Evidence from multiple laboratories and experimental models, as discussed above, confirms that brain injury stimulates neurogenesis within the hippocampus. This augmented proliferation is expressed mainly in the early BrdU+ progenitor cells located in the subgranular layer, although some injury-induced cell death also occurs, affecting mainly immature neurons in the inner granular layer. The present LTP measurements were obtained at 30 dpi and 60 dpi, when the process of integrating TBI-induced newborn GCs, is still ongoing. Evidence shows new dentate GCs require about 3 months to achieve full functional maturity, and up to 6 months for entire integration into the existing network (Zhao et al., 2006; Toni et al., 2007). However, it is during the immature stage when new GCs exhibit enhanced plasticity (Ge et al., 2007, 2008; Schmidt-Hieber et al., 2004; Snyder et al., 2001), and their elevated excitability strongly modulates the sparse coding of the mature GC population (Drew et al., 2016). We previously observed spontaneous recovery of Morris Water Maze performance at 56–60 dpi, but not when maze testing was conducted at 21–25 dpi (Sun et al., 2015). The time course of our observed changes in (−)Pic LTP changes were similar: robust at 60 dpi but deficient at 30 dpi. It is possible that the immature neurons at 30 dpi may represent new neurons born in early post-injury proliferative phase that may not develop normally to support LTP at this time point, whereas at 60 days the new neurons contributing to recovery of LTP may represent cells born at later stage after TBI when the DG may already assume ‘normal’ neurogenesis.

While no causal association can be inferred from this similar sequence, it is quite feasible that a population of cells showing greater plasticity, evident in the adult-born (−)Pic GCs at 60 dpi, would provide functional benefits in the hippocampal processing of a spatial memory task at 56–60 dpi. It is noteworthy that TBI-induced elevation of neurogenesis does not return to baseline until about 2 weeks (Sun et al., 2005), and quantitative estimates of baseline proliferation in the rat dentate gyrus is approximately 9000 new cells each day (Cameron and McKay, 2001). Thus, a large pool of newly-born GCs become available to participate in plasticity and information processing following the injury, and likely contribute to LTP expression and spatial maze performance. The present study was not designed to longitudinally follow a single ‘cohort’ of newly-generated GCs as they integrate, for example cells added by a single day of TBI-induced proliferation. Instead, this investigation examined a population of adult-born GCs, specifically cells that have not yet developed GABAergic inhibition, to test if TBI alters their capacity for LTP and to assess the time frame of any such injury effects. Our results show a dynamic restorative change in the young neurons, and a progressive deterioration in the mature neurons. A greater understanding of TBI-induced functional changes in newborn neurons may lead to strategies for therapeutic interventions against posttraumatic cognitive deficits.

TBI pathology affecting mature granule neurons

In this study, TBI induced a progressive worsening of mature GC LTP plasticity that occurred over the same 60 day time course required for the newly generated GCs to improve their LTP functionality. Other studies using the LFPI model have focused on LTP in other regions of the DG such as CA3 and CA1. A study examining LTP in hippocampal region CA1 found that loss of CA1 LTP was observed at 7 days post-injury following a moderate LFPI in mice, and injury-induced decrease of NMDA potentials and glutamate-mediated excitatory currents may contribute to this LTP changes (Schwarzbach et al., 2006). Similarly, loss of CA1 LTP was reported to be associated to cognitive deficits following a repetitive mild LFPI in rats at 28 days post-injury (Aungst et al., 2014). Other studies examining TBI-induced changes of hippocampal function have reported loss of DG hilar neurons accompanied with abnormal hyperexcitability of GCs at 7 days following LFPI in rats (Lowenstein et al., 1992). A more recent study reported that even following a mild LFPI, the DG demonstrated a diminished capacity to regulate cortical input into the CA3 leading to increased CA3 network excitability (Folweiler et al., 2018). Instances of dissimilar cellular responses to injury, within a single anatomic region, are to be expected in view of the growing appreciation of TBI as a complex pathophysiology with interacting primary and secondary injury cascades (e.g., review by Ng and Lee, 2019).

An interpretation of changes in LTP, observed over a 60 day post-TBI period, should consider results of TBI studies conducted at long survival intervals, well beyond the acute and subacute periods. In our study, we have observed a delayed atrophy of injured hippocampus particularly the DG in our study timeframe. In a more severe LFPI model compare to the present study, significant cortical and hippocampal tissue loss was observed as early as 1 week following TBI and progressive atrophy was observed with time up to one year post-injury (Smith et al., 1997). Another study examining post-TBI brain structural changes at 1 year following a moderate LFPI reported progressive atrophy in white matter (internal capsule) only (Bramlett and Dietrich, 2002). More recent work documented that sustained neuroinflammation accompanied TBI-induced neuronal loss, in measurements extending to 8 months post-injury (Ritzel et al., 2020). It would be expected that LTP recordings which sample from the widest cell population, including the mature GCs which comprise about 90% of the total GC number (Imayoshi et al., 2008), would most accurately reflect the most prevalent and ongoing TBI pathology. Accordingly, the progressive loss of LTP magnitude we observed in the (+)Pic recordings, reduced at 30 dpi and further degraded at 60 dpi, likely represents the chronic neuronal death and cellular dysfunction seen at long survival intervals. Yet despite the clear evidence for ongoing and progressive cellular pathology after TBI, extensive prior research has demonstrated that spontaneous cognitive recovery does occur after these injuries, including work in our lab showing recovery of hippocampal-dependent memory performance during this same 60 day post-TBI interval (Sun et al., 2007, 2015). Research efforts to assist recovery from brain injury, are often designed to facilitate innate restorative mechanisms, including cellular and molecular plasticity (Griesbach and Hovda, 2015; Werner and Stevens, 2015), dynamic brain reorganization / repair (Johnstone et al., 2015; McGinn and Povlishock, 2015), endogenous growth factors (Sun et al., 2009, Weston and Sun, 2018), and extracellular matrix regulation of plasticity (Phillips and Reeves, 2001; Phillips et al., 2014). The present finding, that a spontaneous recovery of the capacity for LTP in adult-born GCs seen in (−)Pic recordings, potentially represents an additional innate restorative mechanism whereby distinctive properties of the young GCs may contribute to behavioral and electrophysiological recovery.

Special properties of adult-born granule cells in plasticity and learning

There is substantial evidence that adult-born GCs play specialized roles in hippocampal function and in specific learning and memory contexts. Blocking neurogenesis impairs performance in hippocampal-dependent learning tasks (Imayoshi et al., 2008), and we have previously shown that the use of an antimitotic agent to transiently inhibit TBI-induced cell proliferation in the DG abolished the innate cognitive recovery of spatial maze performance at 56–60 dpi (Sun et al., 2015). The use of immediate early genes to map neuronal activity has shown newborn GCs are preferentially active during behaviors that involve spatial memory (Kee et al., 2007; Tashiro et al., 2007). The new GCs may be critical for certain information processing tasks before they reach full maturity, with these roles enabled by distinctive physiological properties. Young GCs exhibit a lower threshold for LTP induction (Schmidt-Hieber et al., 2004; Wang et al., 2000), and have a higher expression level of NR2B-containing NMDA receptors (Snyder et al. 2001; Ge et al. 2007). These properties point to an enhanced synaptic plasticity in the newly generated GCs, and to the concept that the young GC population may be preferentially recruited in exploratory behavior and spatial memory tasks. Because of their enhanced responsiveness, adult-born GCs may act to regulate DG processing by modulating the substantially more abundant mature GCs. Providing evidence for this modulatory role was the observation that activation of young adult-born GCs increased GABAergic inhibition onto mature GC, and reduced the number of mature GCs activated by exploration (Drew et al., 2016).

Multiple aspects of the continuously generating new GCs, make them a favorable target population in post-TBI therapies based on drug or growth factor treatments or environmental enrichment. Although the newborn cells comprise a numerical minority of the total GC population, they exert a disproportional influence over DG function. For example, in tracking the integration of adult-generated GCs into memory networks, Kee and colleagues (2007) reported the new cells were more likely than mature GCs to be recruited into circuits supporting spatial memory. Accordingly, therapies to ameliorate TBI-induced spatial memory deficits, may prove to be more efficient if aimed at the young GC population. Neurogenesis in the DG is highly sensitive to experience and environmental factors, which would benefit additional approaches to strategic intervention. Physical activity stimulated a significant increase in the generation of new GCs, and enhanced subsequent spatial memory performance and LTP (van Praag et al., 1999). A growing literature has demonstrated environmental enrichment to be reliably effective as an intervention after experimental TBIs (Bondi et al., 2014), and extensive exposure to an enriched environment was found to significantly increase hippocampal neurogenesis (Kempermann et al., 1997). Regardless, of the intervention approach, treatments to facilitate a restorative contribution of new GCs may benefit from a protracted therapeutic window. Specifically, Lemaire et al (2012) reported that experience in the water maze could induce dendritic plasticity in adult-born GCs even when the cells are several months of age.

Conclusion

This was the first study to assess the effects of TBI on LTP plasticity in adult-born GCs. A profound LTP deficit was initially observed in the newly-generated GCs at 30 dpi, but by 60 dpi the capacity for LTP in this population was restored to control levels. An opposite outcome was observed for the mature GC population, with LTP impairment seen at 30 dpi, and further exacerbated at 60 dpi. The time-frame of LTP improvements in the newborn cell population, similar to that of post-TBI recovery of spatial maze performance found previously, suggests that the unique functional properties of newborn granule cells enable them to contribute to restorative change following brain injury.

Highlights.

• Understanding cellular mechanisms of the role of DG neurogenesis in post-TBI recovery

• Investigating time-dependent changes in the capacity in long-term potentiation (LTP) within the DG in relation to TBI-induced neurogenesis

• Finding of the unique function of newborn granule cells in contributing to restorative change following brain injury.

Abbreviations

GCs

granule cells

DG

dentate gyrus

fEPSPs

evoke field excitatory postsynaptic potentials

H&E

hematoxylin and eosin

HFS

high frequency stimulation

LFPI

lateral fluid percussion injury

LTP

long-term potentiation

MPP

medial perforant pathway

SGZ

subgranular zone

TBI

traumatic brain injury