Abstract
Post-traumatic epilepsy (PTE) and behavioral comorbidities frequently develop after traumatic brain injury (TBI). Aberrant neurogenesis of dentate granule cells (DGCs) after TBI may contribute to the synaptic reorganization that occurs in PTE, but how neurogenesis at different times relative to the injury contributes to feedback inhibition and recurrent excitation in the dentate gyrus is unknown. Thus, we examined whether DGCs born at different postnatal ages differentially participate in feedback inhibition and recurrent excitation in the dentate gyrus using the controlled cortical impact (CCI) model of TBI. Both sexes of transgenic mice expressing channelrhodopsin2 (ChR2) in postnatally born DGCs were used for optogenetic activation of three DGC cohorts: postnatally early born DGCs, or those born just before or after CCI. We performed whole-cell patch-clamp recordings from ChR2-negative, mature DGCs and parvalbumin-expressing basket cells (PVBCs) in hippocampal slices to determine whether optogenetic activation of postnatally born DGCs increases feedback inhibition and/or recurrent excitation in mice 8-10 weeks after CCI and whether PVBCs are targets of ChR2-positive DGCs. In the dentate gyrus ipsilateral to CCI, activation of ChR2-expressing DGCs born before CCI produced increased feedback inhibition in ChR2-negative DGCs and increased excitation in PVBCs compared with those from sham controls. This upregulated feedback inhibition was less prominent in DGCs born early in life or after CCI. Surprisingly, ChR2-positive DGC activation rarely evoked recurrent excitation in mature DGCs from any cohort. These results support that DGC birth date-related increased feedback inhibition in of DGCs may contribute to altered excitability after TBI.
SIGNIFICANCE STATEMENT Dentate granule cells (DGCs) control excitability of the dentate gyrus through synaptic interactions with inhibitory GABAergic interneurons. Persistent changes in DGC synaptic connectivity develop after traumatic brain injury, contributing to hyperexcitability in post-traumatic epilepsy (PTE). However, the impact of DGC neurogenesis on synaptic reorganization, especially on inhibitory circuits, after brain injury is not adequately described. Here, upregulation of feedback inhibition in mature DGCs from male and female mice was associated with increased excitation of parvalbumin-expressing basket cells by postnatally born DGCs, providing novel insights into underlying mechanisms of altered excitability after brain injury. A better understanding of these inhibitory circuit changes can help formulate hypotheses for development of novel, evidence-based treatments for post-traumatic epilepsy by targeting birth date-specific subsets of DGCs.
- adult neurogenesis
- mossy fiber sprouting
- optogenetics
- parvalbumin-expressing basket cell
- patch-clamp
- post-traumatic epilepsy
Introduction
Traumatic brain injury (TBI) refers to a disruption of brain function caused by physical impact to the head. Post-traumatic epilepsy (PTE) is often intractable to current medical treatment and is a major complication of TBI (Annegers et al., 1998; Lowenstein, 2009; Raymont et al., 2010; Smith, 2016). Functional reorganization of neuronal circuits in the dentate gyrus develops after TBI (Hunt et al., 2013; Ngwenya and Danzer, 2019; Redell et al., 2020), and GABAergic interneurons and dentate granule cells (DGCs) are critically implicated in that reorganization (Santhakumar et al., 2001; Hunt et al., 2009, 2010, 2011; Ibrahim et al., 2016; Butler et al., 2017; Frankowski et al., 2019; Zhu et al., 2019). DGCs are generated throughout the lifespan by neurogenesis (Schlessinger et al., 1975; Jones et al., 2003; Kronenberg et al., 2006; Mathews et al., 2010; Zhao et al., 2010; Lopez-Rojas and Kreutz, 2016). Importantly, neuronal excitability of the dentate gyrus is modulated by integrating adult born DGCs with feedback inhibitory circuits (Ikrar et al., 2013; Temprana et al., 2015; Drew et al., 2016), whereas the aberrant adult neurogenesis that occurs transiently after TBI may contribute to hyperexcitability (Butler et al., 2015; Neuberger et al., 2017). Specifically, DGCs manifest sprouted mossy fibers in the inner molecular layer that form recurrent excitatory synapses on other DGCs, which may contribute to recurrent excitation after TBI (Golarai et al., 2001; Santhakumar et al., 2001; Hunt et al., 2009; Butler et al., 2015). Conversely, GABAergic interneurons in the dentate gyrus and hilus may be a major target of sprouted mossy fibers in PTE and temporal lobe epilepsy (TLE) (Sloviter, 1992; Santhakumar et al., 2001; Sloviter et al., 2006; Hunt et al., 2011). Thus, sprouted mossy fibers may influence dentate gyrus excitability and spontaneous seizures through local GABAergic interneurons. However, whether and how adult born DGCs drive feedback inhibitory circuits is uncertain in PTE, primarily because selective activation of adult born DGCs in the dentate gyrus remains technically challenging. Identifying synaptic targets of adult born DGCs will advance our understanding of reorganization of dentate gyrus circuits after TBI.
Prior studies attempting to uncover the role of adult born DGCs in hyperexcitability and spontaneous seizures in PTE and TLE suggested distinct contributions of DGCs born at different ages to the functional reorganization of local circuits (Hendricks et al., 2017; Neuberger et al., 2017). Long-term tracing studies of birth date-labeled adult born DGCs in a TLE model demonstrated that mature DGCs were least affected by epileptogenic brain insults and remained normal morphologically (Walter et al., 2007; Kron et al., 2010; Althaus et al., 2019). In contrast, DGCs born just before or after epileptogenic brain insults displayed abnormal morphologic properties, including aberrant hilar basal dendrites, mossy fiber sprouting, and prominent ectopic migration (Walter et al., 2007; Kron et al., 2010; Villasana et al., 2015; Ibrahim et al., 2016; Althaus et al., 2019). Recent results from optogenetic analysis of dentate gyrus recurrent excitatory circuits in TLE have also corroborated DGC birth date-differential contribution to the functional reorganization of dentate gyrus neuronal circuits (Hendricks et al., 2017). However, it is unknown how adult born DGCs affect functional synaptic connectivity with GABAergic interneurons and mature DGCs after TBI.
Here, we have targeted three different cohorts of DGCs that were born at different maturation periods (i.e., neonatally born, or born either just before or after TBI) by using the Gli1-CreERT2::Channelrhodopsin-2 (ChR2) mouse strain, which permanently expresses ChR2/EYFP fusion protein in newborn DGCs in an inducible fashion. We focused on the following questions: (1) Do postnatally born DGCs at different ages differentially contribute to feedback inhibition and recurrent excitation several weeks after experimental TBI? (2) What are circuit mechanisms underlying increased feedback inhibition in the dentate gyrus after TBI? (3) Are postnatally born DGCs at different ages differentially vulnerable to TBI?
Materials and Methods
Animals
All procedures involving animals used in this study were approved by the Institutional Animal Care and Use Committees of the University of Kentucky and Colorado State University. All mice were maintained at ambient temperature and humidity under a 14 h light/10 h dark cycle. Standard chow and water were available ad libitum. For the current study, we crossed hemizygous Gli1-CreERT2 strain (stock #007913, The Jackson Laboratory) to homozygous Ai32 mouse strain that expresses a ChR2/EYFP fusion protein in a Cre recombinase-dependent manner (stock #024109, The Jackson Laboratory) and produced Gli1-CreERT2::ChR2 mice. The double-transgenic mice were intraperitoneally injected with tamoxifen (2 mg and 127.5 µg for adults and neonatal pups per injection, respectively; T5648, Millipore Sigma) every other day during one of three periods (Fig. 1A): postnatal day 8-12 (three injections; for labeling DGCs that were mature when brain injury was induced; termed “early born”), 5-7 weeks of age (seven injections; for labeling DGCs born just before controlled cortical impact [CCI] to identify DGCs that were immature when brain injury was induced; termed “adult born pre-CCI”), 7-9 weeks of age (seven injections; for labeling DGCs born just after CCI to identify DGCs born just after brain injury; termed “adult born post-CCI”). The rate of neurogenesis is significantly higher at perinatal and neonatal periods, and these early born DGCs numerically dominate adult DGCs (Muramatsu et al., 2007), so tamoxifen treatment to label early born DGCs was less to reduce the disparity in cell numbers between the groups. The last dose of tamoxifen was injected 2 d before CCI injury for the “adult born pre-CCI” group, and the first dose of tamoxifen was injected 2 d after CCI injury for “adult born post-CCI” group. Tamoxifen administration resulted in persistent expression of ChR2/EYFP in neural progenitor cells and their progeny in the hippocampus, as previously reported (Fig. 1B,C,C′) (Ahn and Joyner, 2005; Hester and Danzer, 2013).
CCI model of TBI
Seven-week-old Gli1-CreERT2::ChR2 mice were subjected to CCI to induce unilateral focal injury, as previously described (Figs. 1A, 2) (Hunt et al., 2009, 2010, 2011; Butler et al., 2015, 2016, 2017). Equal numbers of male and female mice from each of the tamoxifen-treated groups were subjected to unilateral TBI or craniotomy (sham control). Briefly, the head of an isoflurane-anesthetized mouse was fixed by the ear bars of the stereotaxic instrument (David Kopf Instruments), and a round 4 mm craniotomy was made between lambda and bregma on the left skull, exposing the frontoparietal lobe. Craniotomy was performed using a low-speed drill, and the bone flap was carefully removed using Dumont forceps without breaching the dura. Severe cortical contusion injury was induced via a 3 mm beveled impactor tip that delivered a pneumatic impact (3.5 m/s velocity, 1.0 mm penetration depth, 500 ms dwell time) using a head trauma contusion device (TBI-0310, Precision Systems and Instrumentation). A small piece of Surgicel (Johnson & Johnson) was placed over the injured area, and the scalp incision was repaired by using 5–0 sutures. This procedure resulted in cortical cavitation and hippocampal distortion along with discontinuous, punctate mossy fiber sprouting in the inner molecular layer (Fig. 2), and excitatory synaptic reorganization ipsilateral to CCI injury, as previously reported (Hunt et al., 2009, 2010; Butler et al., 2015). Similar CCI injury develops PTE with spontaneous seizures in ∼40% of mice by 8-10 weeks after injury (Hunt et al., 2009). Craniotomy alone was performed on age-matched mice as sham controls. All postsurgical mice were individually housed and recovered in the vivarium. Survival rate for CCI- and sham-injured mice was 100%.
Whole-cell patch-clamp recordings
Coronal slices (300 μm) of the dorsal hippocampus (anterior–posterior: ∼−1.34 to −2.54 mm from bregma; four ipsilateral and four contralateral slices obtained from each mouse) were prepared using a vibrating microtome (Vibratome Series 1000; Technical Products) from both sexes of Gli1-CreERT2::ChR2 mice 8-10 weeks after CCI or sham injury. This experimental timeline was selected to standardize mouse age and the time after injury. Immediately after cutting, hippocampal slices were incubated in sucrose-containing artificial cerebrospinal fluid (ACSF) for 1 h at 33°C, and then stored at room temperature until electrical recording. Sucrose containing ACSF consisted of the following (in mm): 85 NaCl, 75 sucrose, 25 glucose, 24 NaHCO3, 4 MgCl2, 2.5 KCl, 1.25 NaH2PO4, and 0.5 CaCl2. Alternate brain slices were assigned for either electrophysiology or histology. For immunohistochemistry, slices ipsilateral and contralateral to CCI (two each side) were immediately fixed in a solution containing 4% paraformaldehyde (PFA) and 0.2% picric acid in 0.1 m phosphate buffer (PB) pH 7.4, and stored at 4°C for 24-48 h. Hippocampal slices destined for electrophysiology were transferred to a recording chamber in ACSF containing the following (in mm): 124 NaCl, 26 NaHCO3, 11 glucose, 3 KCl, 2 CaCl2, 1.3 MgCl2, and 1.4 NaH2PO4 for most electrical recordings. We used Mg2+-free ACSF containing 30 μm bicuculline to induce a hyperexcitable condition in a subset of electrical recordings, where indicated. Hippocampal slices were visualized on an upright microscope (BX51WI, Olympus) with infrared-differential interference contrast optics. Optogenetic stimulation was performed using LED (Thorlabs DC2100) or xenon lamp light sources (LAMBDA XL, Sutter Instrument).
To examine ChR2-positive DGC-mediated feedback inhibition and recurrent excitation, whole-cell patch-clamp recordings from ChR2-negative DGCs that were located within the outer one-half of the granule cell layer (GCL) were performed in voltage-clamp configuration. The pipette solution contained the following (in mm): 126 K-gluconate, 4 KCl, 10 HEPES, 4 ATP-Mg, 0.3 GTP-Na, 10 phosphocreatine, and 5.4 biocytin, pH 7.2 (osmolarity of 290-295 mOsm and patch pipettes had resistances of 3-5 mΩ). All electrical recordings were made at 32°C-34°C using a MultiClamp 700B amplifier (Molecular Devices). Electrical signals were filtered at 10 kHz using a Bessel filter and digitized at 20 kHz (Digidata 1440A; Molecular Devices). Electrical recordings were discarded if the series resistance changed >20% during a recording or reached 25 mΩ. The recorded traces were analyzed using Clampfit software (Molecular Devices, version 10.7.0) and Mini Analysis (Synaptosoft, version 6.0.7). Blue light (30 ms duration) was delivered through a 40× objective to the dentate gyrus and the hilus (light power density, 9.8-9.9 mW/mm2). The power density was estimated from the FOV (662 µm) of 40× objective (LUMPlanFL N, NA 0.8; Olympus). Blue light was centered on the somata of recorded cells. As we previously described, evoked IPSCs (eIPSCs) were measured at 0 mV (i.e., near the reversal potential of ionotropic glutamate receptor-mediated events) (Butler et al., 2020), and evoked excitatory postsynaptic currents (eEPSCs) were recorded at –70 mV (i.e., near the reversal potential of GABAA receptor-mediated events). Ten light stimuli (30 ms duration, 30 s interval) were used to analyze eIPSCs and eEPSCs in each DGC.
To examine intrinsic properties of PVBCs, whole-cell patch-clamp recordings from PVBCs were performed in current-clamp configuration. Membrane responses of PVBCs were evoked from −60 mV by 1 s current steps (−200 to 500 pA, 50 pA increments). To examine PVBC excitation by ChR2-positive DGCs, three types of whole-cell patch-clamp recordings from PVBCs were performed. In voltage-clamp configuration, eEPSCs in PVBCs (Vhold = −70 mV) evoked by 30 ms light stimuli (10 stimuli, 30 s interval) were measured. Additionally, three trains of eEPSCs in PVBCs were evoked by 30 ms light stimuli (10 light stimuli at 10 Hz) every minute to characterize short-term plasticity of network-driven synaptic excitation. Finally, we recorded in current-clamp configuration to examine whether optogenetic activation of ChR2-positive DGCs produced action potentials (APs) in PVBCs. Specifically, subthreshold and suprathreshold membrane responses in PVBCs were evoked from −60 mV by 30 ms light stimuli (10 stimuli, 30 s interval).
The properties of eIPSCs and eEPSCs in ChR2-negative DGCs were measured as we previously described (Hunt et al., 2010; Kang et al., 2021). IPSCs and EPSCs were manually detected using Mini Analysis (Synaptosoft). Only currents with amplitudes more than 3× the root mean square noise level were included for analysis. IPSC and EPSC frequency was measured in 100 ms bins for 500 ms before and after each light stimulus, and responses were averaged for each cell. Evoked inhibitory and/or excitatory connections were considered to be present if the following two criteria were satisfied: (1) the number of EPSCs/IPSCs in the first 100 ms segment after light stimulation was greater than the mean number of events/100 ms before light stimulation + 3× SDs; and (2) a response was observed in at least 30% of trials. For PVBCs, eEPSC amplitude was measured from average current before stimulation to peak in 100 ms segment from light stimulus onset. Charge transfer of eEPSCs in PVBCs was measured from 100 ms current response from light stimulus onset and averaged for each cell. The duration of eEPSCs in PVBCs was measured from the peak of the first event to the peak of the last event.
Intrinsic properties of PVBCs were measured as we previously described (Kang et al., 2018, 2021). The following properties were examined: (1) resting membrane potential was measured from the average voltage after an equilibration period (>2 min); (2) input resistance was measured from voltage responses to 1 s current steps (−50, 0, and 50 pA). The slope of current–voltage curve was calculated from the membrane potential at 0.8-1.0 s from the start of current steps; (3) membrane time constant was measured from voltage responses to 1 s, −50 pA steps; single exponential functions were fitted to the voltage responses to the start of current step to the hyperpolarization peak to measure decay time constant; and (4) AP threshold, amplitude, and half-width were measured from the first AP evoked by rheobase. AP threshold was defined as the point at which the derivative of the membrane potential (dV/dt) first exceeded its mean by 2 SDs. AP amplitude was measured from threshold to peak; half-width was defined as the AP duration at one-half AP amplitude.
PVBC identification
All PVBCs were identified based on post hoc anatomic properties (i.e., basket-like axonal morphology predominantly located in the GCL) and expression of PV (Lee et al., 2010, 2014, 2015, 2017; Lee and Soltesz, 2011; Kang et al., 2018; Elgueta and Bartos, 2019; Vaden et al., 2020). All tested interneurons (10 for sham control and 7 for CCI) manifested basket-like axon terminals in the GCL along with immunopositivity for PV, as previously reported (Lee et al., 2010; Lee and Soltesz, 2011; Savanthrapadian et al., 2014; Kang et al., 2018). Briefly, interneurons were first selected under infrared-differential interference contrast based on soma location near the GCL, large cell bodies and multipolar dendritic shapes, and fast spiking firing pattern. After electrical recordings with pipettes containing biocytin, the slices were immersed in 4% PFA and stored at 4°C for 24-48 h. The slices were washed in PB and cryopreserved with 30% sucrose solution. They were embedded and frozen in OCT compound (Thermo Fisher Scientific) and cryosectioned using a cryostat (HM 550; Microm) at 40 μm thickness. Biocytin-filled neurons were stained with streptavidin conjugated with Rhodamine Red-X (Table 1) and first examined for neuronal morphology using an epifluorescence upright microscope, BX43 (Olympus). Z-stack images (319.45 µm × 319.45 µm, 1 µm step interval) were acquired under a 20× objective (Plan-Apochromat, NA 0.8) using a Zeiss LSM800 confocal microscope (Carl Zeiss). Using ImageJ, Z-stack images were projected into two dimensions and merged to assess the morphology of biocytin-filled neurons. The sections were used for immunohistochemistry to examine whether biocytin-filled cells expressed PV. Tissues were blocked in 5% normal donkey serum in Tris-buffered saline (TBS) pH 7.4, containing 0.3% Triton-X 100 for 1 h at room temperature and incubated in primary and secondary antibodies (Table 1) diluted in TBS with 0.3% Triton-X 100 and 1% normal donkey serum. After washing thoroughly in TBS with 0.05% Tween-20, the sections were mounted in Vectashield medium (H-1000, Vector Laboratories) and subjected to Z-stack image acquisitions (159.72 µm × 159.72 µm, 1 µm step interval) using a Zeiss LSM 800 confocal microscope equipped with 40× objective lens (EC Plan-Neofluar, Oil NA 1.3, Carl Zeiss).
Reagents used for fluorescence microscopya
Intrinsic properties of PVBCs in sham control and CCI injured micea
Immunohistochemistry and quantification of adult born pre- and post-CCI and early born DGCs
Fixed brain slices were embedded in OCT compound and sectioned at 40 µm thickness using a cryostat (HM 550; Microm). The sections were blocked in 5% normal goat serum in TBS, pH 7.4, containing 0.3% Triton-X 100 for 1 h at room temperature and incubated in solution containing primary or secondary antibodies diluted in TBS with 0.3% Triton-X 100 and 1% normal goat serum. To quantify the number of mature adult born and early born DGCs around the dentate GCL, tissues were stained with antibodies targeting calbindin D-28K and EYFP (Table 1). Calbindin was selected as a marker of mature DGCs (Kernie et al., 2001; Iwano et al., 2012; Redell et al., 2020). Although the level of calbindin expression varies among mature DGCs (Iwano et al., 2012), this heterogeneity of calbindin expression did not prevent quantification of mature DGCs (Kempermann et al., 1997). We avoided using Prox1, one of the most commonly used DGC-selective markers, since it is expressed in both immature and mature DGCs rather than exclusively expressed in mature DGCs, as is calbindin (Lavado et al., 2010; Iwano et al., 2012). After each incubation in solution containing primary or secondary antibodies, sections were washed thoroughly in TBS containing 0.05% Tween-20 and mounted in Vectashield medium for fluorescence microscopy. The GCLs were imaged from five subregions (i.e., apex, medial and lateral subregions of the upper and the lower blade) for each section of the ipsilateral and contralateral dentate gyrus. Z-stack images (317.4 µm × 317.4 µm, 4.2 µm Z-range with 0.3 µm step interval) were acquired from each subregion using Nikon C2+ Eclipse-Ti confocal microscope (Nikon Instruments) with 40× oil objective (Nikon S FL, NA 1.3). To quantify the number of calbindin/EYFP double-positive mature DGCs, Z-stack image files were randomized and blinded using Microsoft Excel. The number of calbindin/EYFP double-positive cells were counted and normalized by the total number of nuclei of calbindin-positive DGCs within the GCL for each Z-stack using ImageJ (version 1.53c). Calbindin-positive DGC nuclei were quantified by ImageJ (“Analyze Particles”) using DAPI-labeled images of each Z-stack. The mean of calbindin/EYFP double-positive cell densities within each animal was used to test whether there were differences in the cell numbers between sham control and CCI mice.
TIMM staining
To visualize zinc-containing axons of DGCs, we processed brain slices with a TIMM staining method (Winokur et al., 2004). Brain slices were treated with 0.37% Na2S in 0.1 m NaH2PO4 solution, pH 7.2, for 20 min at room temperature, rinsed with PB and fixed in 4% PFA in PB overnight at 4°C. Fixed brain slices were rinsed with PB, cryoprotected in 30% sucrose in PB, embedded in OCT, and sectioned at 40 µm using a cryostat. Sections were mounted on gelatin-coated glass slides and dried overnight. All glassware used in the staining process were washed in 1% HCl solution. Slide-mounted sections were placed in a TIMM developer solution consisting of 170 ml 50% gum arabic, 30 ml of 2.25 m citrate buffer, 50 ml of 0.73% Ag-lactate, and 45 ml of 0.5 m hydroquinone. Brain sections were stained in the dark for 60-70 min at room temperature and washed in gently running tap water, rinsed in deionized water. TIMM-stained sections were lightly counterstained with cresyl violet to stain Nissl bodies and coverslipped with Permount mounting medium. Brightfield images of the hippocampi and the dentate gyri were taken using SPOT system (RT Slider camera model 2.3.1, SPOT Advanced Software version 4.7; SPOT Diagnostic Instruments) and an upright microscope (BX43, Olympus) equipped with 4× and 40× objectives.
Statistics
For most experiments, one-way or two-way ANOVAs were followed by Tukey or Dunn's tests (see Figs. 3, 4, 6, 7, and 9). Fisher's exact test was used to assess the proportion of DGCs showing eIPSCs/eEPSCs and proportion of PVBCs showing evoked APs between sham control and CCI mice without (for two groups; see Fig. 8B) and with Bonferroni–Holm corrections (for three groups; see Figs. 4 D,H,L and 9D,G,J,M-O). Unpaired two-tailed Student's t tests were used to compare two groups when the data showed a normal distribution based on the Shapiro–Wilk test (see Fig. 7; Table 2). If the data were not normally distributed, a Mann–Whitney Rank Sum test (for unpaired data) was used (see Fig. 7; Table 2). Paired two-tailed Student's tests were used for Figure 5. Data are mean ± SEM. A p value <0.05 was considered significant. Statistical analyses were performed using SigmaPlot 14 (Systat Software) or Prism 9 (GraphPad Software).
Results
The long-term impact of CCI injury on the number of early born and adult born DGCs
Early born and adult born DGCs are differentially vulnerable in CCI injury (Wang et al., 2016). However, it is uncertain whether CCI injury differently affects the survival of these distinct cohorts of DGCs 8-10 weeks after injury. We therefore examined whether CCI injury differentially produces a long-term change in the number of early born and adult born DGCs. Postnatally born DGCs were selectively labeled with ChR2/EYFP in sham control and CCI-injured mice (Fig. 1A,B). Since adult born DGC-mediated circuit activity increases with maturation (Temprana et al., 2015) and calbindin is commonly used as a mature DGC marker (Kernie et al., 2001; Iwano et al., 2012; Redell et al., 2020), we quantified the number of EYFP/calbindin double-positive DGCs in sham control and CCI injured mice 8-10 week after CCI injury.
Experimental design of birth date-based expression of ChR2/EYFP in early born and adult born DGCs and light-evoked APs in a ChR2-positive DGC. A, To label early born or adult born DGCs with ChR2, Gli1-CreERT2::ChR2 double-transgenic mice were produced and injected with tamoxifen (i.p.) at one of three periods, postnatal 8-12 d, postnatal 5-7 weeks, and postnatal 7-9 weeks. All mice underwent either CCI injury or craniotomy (sham control) at 7 weeks of age and were killed 8-10 weeks after surgery for electrophysiological and immunohistochemical experiments. B, Dentate gyri ipsilateral to sham surgery expressing ChR2/EYFP in early born or adult born (before and after surgery) DGCs after injecting tamoxifen during one of three periods shown in A. Green and blue represent EYFP and DAPI staining, respectively. C, C′, ChR2/EYFP-expressing adult born DGC recovered after filling with biocytin during a patch clamp recording, as in D and E. Biocytin-filled cell DGC morphology (asterisk indicates soma; arrowheads indicate dendrites; arrow indicates axon) was analyzed post hoc. C, Biocytin-filled DGC (red). C′, The same DGC was EYFP labeled. D, Firing properties of a DGC from a sham control mouse expressing ChR2/EYFP. The DGC was held at resting membrane potential (–82 mV) before it produced accommodating AP discharges in response to 1 s depolarizing current steps (100 and 200 pA). E, Light-evoked APs in the DGC shown in D. Thirty millisecond light pulses delivered every 30 s through the 40× objective reliably produced APs in the DGC; the number of evoked APs was similar between DGCs from sham and CCI injured mice. Blue lines indicate when the blue light was delivered.
EYFP/calbindin double-positive DGCs in sham control and CCI mice were quantified by an investigator blinded to animal treatment. Imaging revealed that the proportion of EYFP/calbindin double-positive early born DGCs was numerically, but not statistically, smaller in the dentate gyrus both ipsilateral and contralateral to CCI injury compared with sham controls (Fig. 3A,D; sham controls: 18.06 ± 3.93%, n = 6; CCI ipsilateral: 9.21 ± 2.41%, n = 6; CCI contralateral: 7.73 ± 2.13%, n = 6; one-way ANOVA, F(2,15) = 3.636, p = 0.052). The p value of 0.052 was considered a statistical trend that may have reached significance with larger sample size (estimated sample size required = 59 animals per group; post hoc power analysis of α = 0.05, 80% power, and estimated mean difference of 20%). These results suggest that a large portion of early born DGCs, which were mature by the time of CCI injury, failed to survive for long periods in the dentate gyrus both ipsilateral and contralateral to CCI injury, which is in agreement with previous findings showing that severe CCI injury mainly caused mature DGC death (Wang et al., 2016). Severe TBI also results in progressive neurodegeneration in the contralateral hemisphere, including the hippocampus and the dentate gyrus, although this hemisphere did not receive mechanical injury directly (Hall et al., 2005). Thus, contralateral neurodegeneration might underlie a reduction in early born DGC number in the dentate gyrus contralateral to CCI injury (Fig. 3A,D).
In contrast, there were similar proportions of EYFP/calbindin double-positive DGCs born just before CCI or sham injury (sham controls: 6.36 ± 1.51%, n = 6; CCI ipsilateral: 6.27 ± 0.52%, n = 6; CCI contralateral: 5.90 ± 0.97%, n = 6; one-way ANOVA, F(2,15) = 0.05061, p = 0.951) and after CCI injury (sham controls: 3.22 ± 0.63%, n = 6; CCI ipsilateral: 1.78 ± 0.43%, n = 6; CCI contralateral: 2.52 ± 0.38%, n = 6; one-way ANOVA, F(2,15) = 2.123, p = 0.154) (Fig. 3B,C,E,F). These results suggest that DGCs that were immature at the time of CCI injury survived and integrated into feedback inhibitory circuits in the dentate gyrus. Interestingly, although adult neurogenesis increases transiently in the 7-14 d after CCI injury (Butler et al., 2015), our results suggest that most DGCs born just after CCI injury were not more likely to survive and integrate into hippocampal circuitry at later times after injury (Fig. 3C,F). These imaging results suggest that DGCs born at different postnatal stages are differentially vulnerable to CCI injury.
CCI increases feedback inhibition in the dentate gyrus
To determine how adult born pre-CCI, adult born post-CCI, and early born DGCs contribute to local circuit operations in the dentate gyrus after brain injury, we first examined whether CCI injury alters feedback inhibition, since GABAergic interneurons in the dentate gyrus and the hilus are major targets of sprouted mossy fibers (Sloviter, 1992; Sloviter et al., 2006). Mossy fiber spouting was evident after CCI injury (Fig. 2), but it is not known how activity of adult born and early born DGCs affects feedback inhibition after TBI. Thus, we sought to examine whether both adult born and early born DGCs contribute to altered GABAergic inhibition in the dentate gyrus using selective expression of ChR2 in a subset of birth dated DGCs and whole-cell patch-clamp recordings from mature DGCs in a mouse model of TBI induced by CCI injury.
Aberrant mossy fiber sprouting in the inner molecular layer after severe CCI injury. TIMM and Nissl-counterstained images near injury epicenter 8-10 weeks after surgery. Dentate gyri ipsilateral to sham control (Aa) and contralateral to CCI (Cc) did not develop structural damages or mossy fiber sprouting in the inner molecular layer. Ipsilateral to CCI, cortical cavitation and hippocampal distortion were observed (B) as was intense mossy fiber sprouting in the inner molecular layer (Bb) near the injury site. Bb, Arrows indicate punctate mossy fibers in the inner molecular layer.
DGCs born just before or after CCI were selectively labeled with ChR2/EYFP in sham control and CCI injured mice (Fig. 1A,B) to examine whether the two cohorts of adult born DGCs differentially participate in the persistent plasticity of GABAergic inhibitory circuits that develops in the dentate gyrus in the weeks following TBI. Light stimulation (30 ms duration) evoked multiple spikes in ChR2-positive DGCs in both sham controls and CCI mice (Fig. 1E). Shorter light pulses (e.g., 15 ms) also produced multiple spikes (n = 3). Therefore, light-evoked multiple spikes in ChR2-positive DGCs are most likely because of intrinsic properties of DGCs (e.g., low-threshold calcium spikes) (Dumenieu et al., 2018) and the slow decay time constant of ChR2-mediated currents (Gunaydin et al., 2010). Importantly, the number of evoked APs was similar between sham controls and CCI mice (sham controls: 2.3 ± 0.2, n = 7; CCI: 2.2 ± 0.3, n = 10; p = 0.688). In addition, there were no direct light-evoked APs in PVBCs (n = 17). These results support selective expression of ChR2 in a subset of DGCs, but not GABAergic interneurons.
Light-evoked activation of the DGCs born just before CCI reliably produced eIPSCs in most ChR2-negative, mature DGCs recorded ipsilateral to CCI at 8-10 weeks after injury (Fig. 4A,D; 29 of 40 DGCs from 6 mice, 72.5%). In sharp contrast, similar light stimuli produced eIPSCs in a significantly smaller proportion of ChR2-negative mature DGCs from sham control mice (Fig. 4C,D; 7 of 59 DGCs from 6 mice, 11.8%; p < 0.001) and contralateral to CCI (Fig. 4B,D; 17 of 42 DGCs from 6 mice, 40.5%; p = 0.004). Furthermore, two-way ANOVA with repeated measures performed on eIPSC number (Fig. 4E) revealed that there was a significant CCI treatment main factor (p = 0.003), time main factor (p < 0.001), and CCI treatment × time interaction (p < 0.001). Specifically, DGCs ipsilateral to CCI injury had significantly more eIPSCs compared with DGCs from sham controls and DGCs contralateral to CCI (Fig. 4E; 100 ms period from light onset; CCI ipsilateral: 1.4 ± 0.1, n = 29; CCI contralateral: 0.8 ± 0.2, n = 17; sham controls: 0.6 ± 0.3, n = 7; p < 0.001 for CCI ipsilateral vs sham controls or CCI contralateral). The amplitude of eIPSCs in ipsilateral DGCs (Fig. 4F; 95.9 ± 15.1 pA, n = 29) was significantly greater compared with sham controls (37.6 ± 4.8 pA, n = 7; p = 0.002) and contralateral to CCI (53.2 ± 7.1 pA, n = 17; p = 0.019). These results suggest that DGCs born just before CCI contribute to persistent upregulation of feedback inhibition of mature DGCs after injury.
We next examined whether optogenetic activation of ipsilateral DGCs born just after CCI also produced significantly greater feedback inhibition in mature DGCs. The proportion of mature DGCs showing eIPSCs was similar between sham control and CCI injured mice (Fig. 4H; sham controls: 7 of 56 DGCs from 6 mice, 12.5%; CCI ipsilateral: 8 of 45 DGCs from 6 mice, 17.7%; CCI contralateral: 4 of 33 DGCs from 6 mice, 12.1%; p = 0.575 or p = 1.0 between sham controls vs CCI ipsilateral or CCI contralateral, respectively). CCI injury also did not result in major changes in eIPSC amplitude or number. Specifically, two-way ANOVA with repeated measures performed on eIPSC number revealed significant time main factor (p < 0.001), but neither for CCI treatment main factor (p = 0.079) nor for CCI treatment × time interaction (p = 0.979). eIPSC amplitude was similar between sham control and CCI injured mice (Fig. 4J; sham controls: 51.3 ± 7.2 pA, n = 7; CCI ipsilateral: 58.3 ± 6.6 pA, n = 8; CCI contralateral: 59.3 ± 13.4 pA, n = 4; p = 0.752). These results suggest that optogenetic activation of DGCs born just after CCI evokes feedback inhibition in mature DGCs that is similar to that observed in sham controls.
We next sought to investigate whether CCI injury also increases early born DGC-mediated feedback inhibition, since early born DGCs participate in functional reorganization of dentate gyrus neuronal circuits that develop after epileptogenic brain insults (Hendricks et al., 2017). The proportion of mature DGCs showing eIPSCs after stimulation of early born DGCs was similar between ipsilateral CCI and sham control brains (Fig. 4L; sham controls: 35 of 53 DGCs from 6 mice, 66.0%; CCI ipsilateral: 33 of 39 DGCs from 6 mice, 84.6%; p = 0.056). Although not statistically significant, the p value of 0.056 was considered a statistical trend toward an increased response probability after CCI. In addition, the proportion of mature DGCs showing eIPSCs was significantly higher in CCI ipsilateral compared with CCI contralateral (Fig. 4L; CCI contralateral: 19 of 37 DGCs from 6 mice, 51.3%; p = 0.002). The number of eIPSCs was similar between sham control and CCI mice (Fig. 4M; time factor, p = 0.001; CCI treatment factor, p = 0.487; time × CCI treatment interaction, p = 0.082; two-way ANOVA with repeated measures). The amplitude of eIPSCs was also similar between sham control and CCI injured mice (Fig. 4N; sham controls: 62.1 ± 6.7 pA, n = 35; CCI ipsilateral: 87.1 ± 11.1 pA, n = 33; CCI contralateral: 64.2 ± 10.1 pA, n = 19; p = 0.181). These results are intriguing since there was a numerical reduction in early born DGCs after CCI (Fig. 3A,D), suggesting that surviving early born DGCs may produce increased excitation of GABAergic interneurons in CCI injured mice compared with sham controls. Additionally, it is likely that early born DGCs ipsilateral to CCI injury increase excitation of GABAergic interneurons, relative to those contralateral to CCI injury, since there was no difference in the number of EYFP/calbindin double-positive, early born DGCs ipsilateral versus contralateral to CCI injury, whereas the proportion of mature DGCs showing eIPSCs was significantly higher ipsilateral versus contralateral to CCI injury.
The proportion of ChR2/EYFP-expressing DGCs was not greater in the dentate gyrus of CCI mice compared with those of sham control mice. A–C, The dentate GCLs of early born (A) and adult born DGC cohorts (B, C, for pre- and post-CCI, respectively) were labeled for nuclei (blue), EYFP (green), and calbindin (red). D–F, Quantification of EYFP/calbindin double-positive DGCs. The number of EYFP/calbindin double-positive DGCs was normalized by the number of calbindin-positive nuclei within GCL. A trend of reductions after CCI was observed for early born EYFP-expressing DGCs in the dentate gyrus ipsilateral and contralateral to CCI injury compared with those of sham controls (A,D; one-way ANOVA, p = 0.052). Proportions of EYFP-expressing DGCs born around the time of injury were not statistically different (B,C,E,F). Number in each bar graph indicates the number of mice used for the study. Data are mean ± SEM.
To account for potential effects of adult born DGC loss on feedback inhibition after CCI, we normalized response strength to the EYFP/ChR2-expressing DGC counts. Specifically, we first multiplied the proportion of DGCs with eIPSCs by mean eIPSC amplitude and then divided that result by the mean number of mature EYFP/ChR2-expressing DGCs for each cohort to obtain an estimate of relative connectivity strength (Fig. 4G,K,O). After controlling for injury-induced DGC count differences, estimated feedback inhibition was upregulated by 3- to 4-fold in both early born and adult born post-CCI cohorts (Fig. 4K,O). Importantly, the upregulation of feedback inhibition in CCI mice for adult born pre-CCI cohort was greater (∼16-fold) than those in early born and adult born post-CCI cohorts (Fig. 4G). Collectively, DGCs that are immature at the time of CCI injury, early born DGCs, and DGCs born just after CCI differentially contribute to the sustained aberrant feedback inhibitory circuits that develop after brain injury, with the adult born pre-CCI DGC cohort being more likely to evoke increased feedback inhibition after CCI injury, compared with those in early born and adult born post-CCI cohorts.
Activation of DGCs born just before CCI, compared with DGCs born after CCI and early born DGCs, produces prominent synaptic inhibition of adult DGCs. A–C, Examples of responses in mature DGCs to optogenetic stimulation of adult born DGCs illustrating response variability in cells from different mice. Single light pulses (blue lines, 30 ms duration every 30 s) were used to induce APs in adult born DGCs, resulting in eIPSCs in recordings from ChR2-negative, mature DGCs. Five consecutive current traces are shown for three DGCs ipsilateral to CCI (A1–A3), CCI contralateral (B), and two DGCs from sham controls (C1,C2) after optogenetic stimulation of DGCs born just before CCI or sham surgery; all sets of traces from different mice. Inset, Expanded time scale of the top trace in A1 (dashed box) displays two eIPSC peaks. D–O, Summary of the proportion of DGCs showing eIPSCs, number of eIPSCs, eIPSC amplitude, and fold increase in feedback inhibition strength for DGCs born just before CCI (D–G), DGCs born just after CCI (H–K), and early born DGCs (L–O). The numbers above the bars indicate number of mature DGCs showing eIPSCs/number of tested mature DGCs (D,H,L) or number of mature DGCs included in eIPSC amplitude measurement (F,J,N). The numbers next to solid circles indicate number of mature DGCs included in IPSC frequency measurement (E,I,M). The relative strength of increase in feedback inhibition, based on the data from Figures 3 and 4, is shown in G, K, O, where the bar graphs represent eIPSC connectivity strength as a function of the mean mature EYFP-expressing DGC number for each cohort and normalized to that of sham controls. Data are mean ± SEM. **p < 0.01; ***p < 0.001; ipsilateral or contralateral to CCI versus sham controls. #p < 0.05; ##p < 0.01; ###p < 0.001; ipsilateral versus contralateral to CCI.
Circuit mechanisms of eIPSCs in the dentate gyrus
Gli1 is selectively expressed in neural precursors that produce DGCs (Ahn and Joyner, 2005; Singh et al., 2015), but not in GABAergic interneurons in the dentate gyrus, so eIPSCs recorded in DGCs are likely because of GABAergic interneuron activation by ChR2-positive DGCs. As expected, bath application of GABAA receptor antagonists (10 μm SR95531 or 30 μm bicuculline) significantly reduced the amplitude of eIPSCs (Fig. 5A,B; predrug control: 100.7 ± 21.3 pA; n = 7; GABAA receptor antagonist: 1.7 ± 1.4 pA, n = 7; p = 0.016). Additionally, application of the non-NMDA ionotropic glutamate receptor antagonist (20 μm DNQX) significantly reduced the amplitude of eIPSCs (Fig. 5C,D; predrug control: 106.9 ± 24.3 pA, n = 11; DNQX: 0.8 ± 0.5 pA, n = 11; p < 0.001), suggesting that eIPSCs in mature DGCs were mediated by activity of local GABAergic interneurons that was induced by stimulation of ChR2-positive DGCs. If ChR2 was also expressed in GABAergic interneurons, DNQX would not completely block eIPSCs. The results are consistent with expression of ChR2 in a subset of DGCs, but not GABAergic interneurons in Gli1-CreERT2::ChR2 double-transgenic mice after tamoxifen injection.
Excitation of GABAergic interneurons by ChR2-positive DGCs is necessary for eIPSCs. A, C, Current traces represent eIPSCs in mature DGCs before and during bath application of a GABAA receptor antagonist, SR95531 (10 μm), or a non-NMDA receptor antagonist, DNQX (20 μm). B, D, Summary of the changes in eIPSC amplitude by GABAA receptor antagonist and non-NMDA receptor antagonist. SR95531 and bicuculline (30 μm) were used in 3 and 4 mature DGCs, respectively, and the results from these 7 DGCs were combined. Replicates are DGC number. Open circles and black solid circles represent the results from individual cells and averages, respectively. Data are mean ± SEM. *p < 0.05. ***p < 0.001.
We further assessed the possibility that these eIPSCs were polysynaptic based on the delay of eIPSC onset from light-evoked APs in ChR2-positive DGCs, which would be expected to be longer than one synaptic delay (1-4 ms) (Hefft and Jonas, 2005; Zhang et al., 2009; Lee et al., 2010). We assessed the latency to (1) the peak of evoked APs in ChR2-positive DGCs and (2) the onset of eIPSCs in ChR2-negative DGCs from light stimulation onset. Based on these two latencies, we measured the latency to eIPSC onset from the peak of evoked APs in ChR2-postive DGCs. As expected, the latency to eIPSC onset from the peak of evoked APs was longer than one synaptic delay in both sham control and CCI injured mice (early born DGC groups: 8.7 ± 0.6 ms, n = 80; before CCI DGC groups: 6.0 ± 0.4 ms, n = 53; after CCI DGC groups: 9.9 ± 1.1 ms, n = 19). Collectively, these results are consistent with the hypothesis that eIPSCs arise from polysynaptic, feedback inhibitory circuits in the dentate gyrus.
Excitation of parvalbumin+ basket cells by adult born DGCs is upregulated in TBI
PVBCs are a major type of GABAergic interneuron in the dentate gyrus that control neuronal excitability and participate in coordinated network oscillations (Hu et al., 2014). PVBC-mediated feedback inhibition is a critical network function for regulation of neuronal excitability in the dentate gyrus (Espinoza et al., 2018). Furthermore, PV+ interneurons are densely innervated by sprouted mossy fibers in humans with TLE and animal models of TLE (Sloviter et al., 2006; Puhahn-Schmeiser et al., 2021). Thus, we hypothesized that PVBCs would be a target for the formation of aberrant feedback inhibition by adult born DGCs after CCI.
To test this hypothesis, we recorded from PVBCs in the dentate gyrus from sham control and CCI mice. We first examined whether there were major changes in intrinsic properties of PVBCs in TBI (Fig. 6). PVBCs in the dentate gyrus from sham control mice manifested high-frequency APs during current injection (Fig. 6B,C). In sharp contrast, PVBCs in the dentate gyrus ipsilateral to CCI injury manifested reduced frequency of evoked APs compared with sham controls (Fig. 6B,C; two-way ANOVA with repeated measures; significant CCI treatment main factor, p = 0.016; significant CCI treatment × injected current amplitude interaction, p < 0.001). Rin of PVBCs after CCI injury was reduced relative to sham controls (Table 2; CCI: 75.5 ± 9.2 mΩ, n = 7; Sham control: 107.3 ± 8.1 mΩ, n = 10; p = 0.022). As expected, PVBCs after CCI injury manifested numerically larger rheobase compared with sham controls, but the difference did not reach statistical significance (Table 2; CCI: 435.7 ± 91.1 pA, n = 7; Sham control: 255.0 ± 33.7 pA, n = 10; p = 0.051). We further examined whether there were major changes in other AP properties of PVBCs. There were no major changes in AP threshold, amplitude, or half-width of PVBCs after CCI injury (Table 2). These results showed that PVBCs manifested reduced firing frequency after CCI injury along with reduced Rin.
CCI-injured mice show reduced firing in PVBCs in the dentate gyrus. A, A representative image of a PVBC. The biocytin-filled PVBC was imaged using confocal microscope. Axon terminals were localized in the GCL of the dentate gyrus, whereas dendrites extended into the inner molecular layer (IML), outer molecular layer (OML), and hilus. The biocytin-labeled PVBC (red) also expressed PV (purple). B, AP firing properties of PVBCs. Evoked APs and voltage responses in PVBCs from CCI and sham control mice during 1 s depolarizing (400 pA) and hyperpolarizing (−200 pA) current steps. There is reduced AP firing in the PVBC from a CCI-injured mouse. C, The summary of AP firing frequency of PVBCs from CCI and sham control mice. n indicates number of PVBCs. Data are mean ± SEM. *p < 0.05. **p < 0.01. ***p < 0.001.
We next examined whether synaptic activation of PVBCs mediated by adult born DGCs was upregulated ipsilateral to CCI injury. We focused on the DGCs born just before CCI injury, since this population of adult born DGCs was the most likely to contribute to increased inhibitory feedback circuits after brain injury (Fig. 4). Optogenetic stimulation of the DGCs born just before CCI injury produced sustained eEPSCs with multiple peaks in both sham controls and CCI injured mice, suggesting that the eEPSCs manifest network driven activity in addition to monosynaptic events (Fig. 7A). Light activation of the DGCs produced greater excitation of PVBCs from CCI mice compared with sham controls. Specifically, peak amplitudes of eEPSCs in PVBCs were significantly larger in CCI mice compared with sham controls (Fig. 7A,B; CCI: 843.6 ± 207.4 pA, n = 8; Sham controls: 94.8 ± 24.4 pA, n = 9; p < 0.001). Charge transfer of eEPSCs was also greater in CCI mice than sham controls (Fig. 7C; CCI: 13.2 ± 3.5 pC, n = 8; Sham controls: 1.5 ± 0.4 pA, n = 9; p = 0.012). In addition, eEPSC bursts were longer in PVBCs after TBI compared with sham controls (eEPSC duration: CCI: 36.2 ± 3.9 ms, n = 8; sham controls: 21.0 ± 4.9 ms, n = 9; p = 0.029).
Activation of DGCs born just before CCI produces greater evoked EPSCs in PVBCs in the dentate gyrus from CCI-injured mice. A, Light-evoked EPSCs in PVBCs in the dentate gyrus from CCI-injured and sham control mice (blue lines, 30 ms duration, 30 s intervals). B, The summary of peak amplitudes of evoked EPSCs and (C) charge transfer. D, eEPSCs in PVBCs in the dentate gyrus from CCI and sham control mice by theta-frequency light stimuli (10 Hz, 10 stimuli, 30 ms duration). E, The summary of peak amplitude as a percent of the first peak amplitude during 10 Hz light stimulations. Data were normalized to first current amplitude in each PVBC. An increase in paired pulse ratio of evoked currents was observed in CCI injured mice versus sham controls. n indicates number of PVBCs. Data are mean ± SEM. *p < 0.05. ***p < 0.001.
Next, we examined whether TBI altered short-term plasticity of eEPSCs using trains of theta-frequency light stimuli. There was an increase in paired pulse ratio of eEPSCs in CCI injured mice compared with sham controls (Fig. 7D,E; two-way ANOVA with repeated measures; significant CCI treatment main factor, p = 0.035; no significant CCI × stimulation number interaction, p = 0.101; no significant stimulation number factor, p = 0.295). Given that sprouted mossy fiber-to-mature DGC synapses in TLE manifest short-term depression (Hendricks et al., 2019), it is interesting that such short-term depression of eEPSCs was not observed in PVBCs from CCI injured mice (Fig. 7E; first amplitude: 842.7 ± 327.5 pA, n = 5; 10th amplitude: 923.1 ± 288.9 pA, n = 5; p = 0.841). These results suggest that the increased synaptic excitation of PVBCs by adult born DGCs after TBI is mainly driven by increased synaptic contacts and/or an injury-associated change in excitatory network activity in the dentate gyrus.
We also examined whether TBI increases AP discharge in PVBCs evoked by adult born DGC stimulation. As illustrated in Figure 8, light activation of the DGCs born just before CCI injury produced AP discharges in a greater portion of PVBCs compared with sham controls (CCI ipsilateral: 4 of 6 PVBCs; Sham control: 1 of 10 PVBCs; p = 0.035). Collectively, these results show that, despite the reduced intrinsic excitability of PVBCs after CCI injury, DGCs born just before CCI injury are more likely to form excitatory synaptic inputs onto PVBCs that are capable of consistently inducing APs in PVBCs that result in increased synaptic inhibition of DGCs.
Activation of the DGCs born just before CCI produces APs in a greater portion of PVBCs in the dentate gyrus after CCI injury. A, Similar light stimuli (30 ms duration, 30 s intervals) produced small depolarization and AP firing in PVBCs in the dentate gyrus from sham control and CCI mice, respectively. Representative 5 light-evoked voltage responses were shown for CCI-injured and sham control mice. B, The summary of proportion of PVBCs showing evoked APs. The numbers (n) above the bars indicate number of PVBCs showing evoked APs/number of tested PVBCs. *p < 0.05.
Neither adult born DGCs nor early born DGCs contribute to recurrent excitation associated with TBI
Sprouted mossy fibers caused by epileptogenic brain insults form functional recurrent excitatory circuits among DGCs by 8-10 weeks after the insult, contributing to hyperexcitability in experimental models of PTE and TLE (Wuarin and Dudek, 2001; Winokur et al., 2004; Hunt et al., 2010; Hendricks et al., 2017), and mossy fiber sprouting was evident after CCI injury here (Fig. 2). However, it is unknown whether early born DGCs and adult born DGCs contribute to recurrent excitation in the dentate gyrus after CCI injury in mice. Therefore, we examined whether recurrent excitation was differentially evoked by optogenetic activation of early born or adult born DGCs.
We first examined whether optogenetic activation of mature DGCs born during the peri-injury period produced recurrent excitation in the dentate gyrus from CCI injured mice. Surprisingly, optogenetic activation of the DGCs born just before or after CCI produced no eEPSCs in ChR2-negative mature DGCs from CCI mice. Specifically, activation of DGCs born just before CCI produced no eEPSCs in ChR2-negative, mature DGCs (Fig. 9A,B,D; CCI ipsilateral: 0 of 41 DGCs from 6 mice; CCI contralateral: 0 of 43 DGCs from 6 mice). Similar results were produced in sham control mice (Fig. 9C,D; sham controls: 1 of 61 DGCs from 6 mice). Accordingly, two-way ANOVA with repeated measures performed on the number of EPSCs revealed not only that there was no significant time main factor (Fig. 9E; p = 0.925), but also that there was no difference in the number of EPSCs between sham control and CCI mice (Fig. 9E; p = 0.721 and p = 0.065 for sham controls vs CCI ipsilateral and CCI contralateral, respectively, post hoc Tukey test). In addition, optogenetic activation of mature DGCs born just before CCI produced no changes in EPSC amplitude in both sham control and CCI injured mice (Fig. 9F; CCI treatment factor, p = 0.056; time factor, p = 0.101; CCI treatment factor × time factor interaction, p = 0.328; two-way ANOVA with repeated measures). These results suggest that aberrant circuit formation of DGCs born just before CCI injury is limited to feedback inhibitory circuits and not recurrent excitatory synapses between DGCs in the weeks following injury.
Neither adult born DGCs nor early born DGCs contribute to recurrent excitation in the dentate gyrus in CCI-injured mice. A–C, Representative examples of current traces indicate no eEPSCs in ChR2-negative mature DGCs. Single light pulses (blue lines, 30 ms duration, every 30 s) were used to produce eEPSCs. Five consecutive current traces are shown for CCI ipsilateral, CCI contralateral, and sham controls of DGCs born just before CCI. D–F, DGCs born just before CCI. G–I, DGCs born just after CCI. J–L, Early born DGCs. Summary of proportion of DGCs showing eEPSCs (D,G,J), EPSC number (E,H,K), and EPSC amplitude (F,I,L). M–O, Summary of proportion of DGCs showing eEPSCs observed in Mg2+-free ACSF containing bicuculline (30 μm). The numbers above the bars indicate number of mature DGCs showing eEPSCs/number of tested mature DGCs. The numbers next to solid circles indicate number of mature DGCs included in EPSC frequency and amplitude measurements. Data are mean ± SEM.
Optogenetic activation of DGCs born just after CCI also produced no eEPSCs in ChR2-negative mature DGCs in sham control and CCI mice (Fig. 9G; CCI ipsilateral: 0 of 46 DGCs from 6 mice; CCI contralateral: 0 of 34 DGCs from 6 mice; sham controls: 0 of 56 DGCs from 6 mice). As expected, two-way ANOVA with repeated measures performed on the number of EPSCs (Fig. 9H) revealed that there was neither significant time main factor (p = 0.949) nor time factor × CCI treatment factor interaction (p = 0.983). There was a small, but significant, reduction in the number of EPSCs from CCI contralateral (Fig. 9H; p = 0.045; post hoc Tukey test), but not from CCI ipsilateral (p = 0.060; post hoc Tukey test) compared with sham controls. In addition, optogenetic activation of mature DGCs born just after CCI did not alter EPSC amplitude in sham control and CCI mice (Fig. 9I; time factor, p = 0.131; time factor × CCI treatment factor interaction, p = 0.834; two-way ANOVA with repeated measures). There was a significant increase in EPSC amplitude from CCI contralateral (Fig. 9I; p = 0.003; post hoc Tukey test), but not from CCI ipsilateral (p = 0.059; post hoc Tukey test) compared with sham controls. These results suggest that adult born DGCs do not contribute significantly to recurrent excitation of DGCs that develops after CCI injury.
We next examined whether early born DGCs contributed to recurrent excitation in CCI injured mice. Optogenetic activation of early born DGCs rarely produced eEPSCs in mature ChR2-negative DGCs in both sham control and CCI mice (Fig. 9J; sham controls: 3 of 54 DGCs from 6 mice, 5.5%; CCI ipsilateral: 3 of 39 DGCs from 6 mice, 7.6%; CCI contralateral: 0 of 38 DGCs from 6 mice, 0%; p = 0.692 and p = 0.265 for CCI ipsilateral and contralateral, respectively, vs sham controls). As expected, there was a significant time main factor, but neither CCI treatment main factor nor CCI treatment factor × time factor interaction for the number of EPSCs (Fig. 9K; time factor: p < 0.001; CCI treatment factor: p = 0.807; CCI treatment factor × time factor interaction: p = 0.136; two-way ANOVA with repeated measures). Optogenetic activation of early born DGCs also changed the amplitude of EPSCs in both sham control and CCI mice (Fig. 9L; p = 0.007, time factor, two-way ANOVA). These results suggest that early born DGCs do not contribute to the increased recurrent excitation of DGCs that develops after CCI injury.
Finally, we sought to examine whether a hyperexcitable condition (e.g., Mg2+-free ACSF containing GABAA receptor blocker, bicuculline) unmasks recurrent excitation that is initiated by activity of postnatally born DGCs. Similar to the results in normal ACSF, optogenetic activation of early born DGCs and adult born DGCs in hippocampal slices perfused with Mg2+-free ACSF containing bicuculline (30 μm) rarely produced eEPSCs in mature DGCs from either sham control or CCI mice (Fig. 9M–O). Collectively, these results showed that early born DGCs and adult born DGCs did not contribute substantially to recurrent excitation among DGCs that develops after CCI injury in mice.
Discussion
This study showed that DGCs born just before CCI injury critically contribute to increased feedback inhibition in the dentate gyrus in a mouse model of TBI. While these results are consistent with prior studies showing increased excitation of hilar interneurons by DGCs and increased inhibition of DGCs by a subset of GABAergic interneurons after epileptogenic brain insults (Santhakumar et al., 2001; Zhang et al., 2009; Hunt et al., 2011), novel aspects of DGC birth date-related upregulation of feedback inhibition after TBI, including increased excitation of PVBCs, are defined. Surprisingly, we found no evidence for participation of postnatally born DGCs in sprouted mossy fiber-mediated recurrent excitation in the dentate gyrus previously reported in the CCI model (Hunt et al., 2009, 2010). These results suggest that GABAergic interneurons, but not DGCs, are major targets of postnatally born DGCs 8-10 weeks after injury and that birth date-specific cohorts of DGCs differentially contribute to dentate gyrus circuit operations via increased feedback inhibition after TBI.
GABAergic PVBCs and hilar perforant path-associated (HIPP) interneurons are reciprocally connected to DGCs through perisomatic and dendritic synapses, respectively, forming feedback inhibitory circuits (Houser, 2007). Adult born DGCs innervate PV+ interneurons (Temprana et al., 2015) and increased excitatory synaptic inputs, with no change in intrinsic firing properties, have been shown in surviving PV+ interneurons in the dentate gyrus up to 14 d after TBI (Nichols et al., 2018; Folweiler et al., 2020). Indeed, we describe an increase in synaptic excitation of PVBCs by the DGCs born just before CCI injury, contributing to increased feedback inhibition after TBI. Firing properties of PVBCs were attenuated in our model, possibly because of the longer time points after injury studied here. DGC-mediated excitation of, and inhibition by, surviving HIPP cells is upregulated in PTE and TLE (Zhang et al., 2009; Halabisky et al., 2010; Hunt et al., 2011; Butler et al., 2017), although specific involvement of postnatally born DGCs in HIPP cell activation is undefined. PVBC axon sprouting could also develop after CCI injury, since axons of PV+ interneurons may sprout in TLE (Christenson Wick et al., 2017; Ábrahám et al., 2020). In addition, activation of surviving mossy cells by DGCs could also contribute to polysynaptic feedback inhibition, since mossy cells manifest local projections to hilar interneurons (Houser et al., 2021).
The most striking difference between DGCs born just before and after CCI is that the former contributed more profoundly to upregulated recurrent synaptic inhibition of DGCs. These results are surprising, since adult neurogenesis increases in the dentate gyrus 3-31 d after TBI (Butler et al., 2015; Villasana et al., 2015; Neuberger et al., 2017; Clark et al., 2021), whereas no increase in mature adult born DGC number was detected 8-10 weeks after injury here. Proper migration and integration of newborn neurons into dentate gyrus circuits are critical for their long-term survival (Kuhn, 2015). Since we found no increase in the number of mature DGCs born just after CCI injury, most DGCs born just after CCI may not have survived long-term, suggesting that increased neurogenesis after CCI injury contributes relatively less to increased feedback inhibition in the dentate gyrus 8-10 weeks after injury. DGCs born after injury might eventually contribute to feedback inhibition at later time points, so it will be important to address this possibility in future studies. Since occurrence of human adult hippocampal neurogenesis has been controversial (Kuhn et al., 2018), careful interpretation of our findings will be required for relevance in adult TBI patients.
It is unknown whether DGC birth date-related upregulation of feedback inhibition develops in TLE in a manner similar to what we observed after CCI. Hendricks et al. (2017) used a doublecortin construct to label ChR2/EYFP in birth dated DGCs in mice with pilocarpine-induced TLE. However, hilar interneurons, including PV+ interneurons, also express doublecortin (Klempin et al., 2011), so all experiments were performed in the presence of GABAA receptor antagonists (Hendricks et al., 2017), obviating analysis of feedback inhibition. Determining the properties of adult born DGC activation of interneurons in TLE may help identify mechanisms in common with TBI and PTE.
Unlike prior studies showing that early born and adult born DGCs contribute to recurrent excitation in TLE (Hendricks et al., 2017, 2019), no postnatally born cell group contributed to recurrent excitation in the dentate gyrus after CCI injury here. This incongruity could arise from differences in pathologic properties between the disease models. The mouse CCI model of PTE manifests a modest degree of mossy fiber sprouting in the inner molecular layer of the dentate gyrus ipsilateral to and near the CCI injury epicenter (Hunt et al., 2009, 2010), whereas the pilocarpine-treated mouse model of TLE (Shibley and Smith, 2002; Hendricks et al., 2017) manifests robust mossy fiber sprouting in the inner molecular layer throughout the dentate gyrus bilaterally. Accordingly, mossy fiber-mediated recurrent excitation develops locally in CCI mice (Hunt et al., 2009, 2010) but is far less robust than in TLE models; this could account for the model-specific differences in projections of adult born DGCs. In rodents, a significant number of DGCs are generated peritnatally (Schlessinger et al., 1975; Altman and Bayer, 1990; Jones et al., 2003; Mathews et al., 2010; Zhao et al., 2010; Lopez-Rojas and Kreutz, 2016) and numerically dominate the DGC population during adulthood (Muramatsu et al., 2007). Furthermore, newborn DGC development decreases dramatically with age (Kronenberg et al., 2006; Muramatsu et al., 2007; Mathews et al., 2010; Zhao et al., 2010; Lopez-Rojas and Kreutz, 2016). Since perinatally born DGCs are more numerous than adult born DGCs, the former may contribute to the recurrent excitation reported after CCI (Hunt et al., 2009, 2010). It is also possible that the sparse mossy fiber sprouting after CCI injury and limited labeling of DGC populations prevented observation of recurrent excitatory activity activated by adult born DGCs. Even so, we observed feedback inhibition mediated by DGCs born just before CCI injury, suggesting that these neurons maintain functional lateral inhibition more readily after CCI injury than DGCs born early in development or after CCI injury. Young (4-6 weeks) adult born DGCs can drive synaptic excitation of mature DGCs normally (Luna et al., 2019), so it will be important to determine whether young adult born DGCs transiently excite mature DGCs at earlier time points after TBI.
It remains unclear whether adult born DGCs contribute to mechanisms that promote or protect from seizures after TBI. Adult neurogenesis in the dentate gyrus is transiently upregulated after TBI, and this increased neurogenesis may cause complex beneficial and pathologic effects on spontaneous seizures and behavioral comorbidities (Neuberger et al., 2017; Ngwenya and Danzer, 2019). However, our current studies showed that DGCs born in the first 2 weeks after TBI contributed relatively modestly to increased synaptic connectivity. Increased feedback inhibition mediated by DGCs born just before CCI injury however, may hold significant functional implications for PTE expression. Given the loss of GABAergic interneurons and reduced IPSC frequency in DGCs after CCI injury (Hunt et al., 2011; Boychuk et al., 2016; Butler et al., 2016; Frankowski et al., 2019), observation of de novo excitatory inputs onto surviving GABAergic interneurons may help compensate for the loss of inhibitory control of DGC activity after TBI. This study demonstrates that DGCs that are immature at the time of injury specifically contribute to this plasticity, at least in part, through increased inputs onto PVBCs, likely because of the rapid axon growth of immature DGCs (Hastings and Gould, 1999; Zhao et al., 2006). Interestingly, diffuse TBI in rodents manifests an increase in IPSC frequency in DGCs (Santhakumar et al., 2001), unlike in the focal CCI model. Injury-specific alterations in inhibitory circuits may therefore play distinct roles in modulating DGC excitability across TBI models.
PV interneurons can promote hippocampal epileptiform activity (Ellender et al., 2014), and GABA release from PV interneurons is increased in epileptic circuits, potentially contributing to epileptic seizures (Hansen et al., 2018). A coordinated increase in feedback inhibitory neuron activity may play a key role in system hyperexcitability and spontaneous seizures in PTE, since upregulated input from GABAergic interneurons can evoke paradoxical rebound APs and synchronize activity in hippocampal circuits (Cobb et al., 1995; Chen et al., 2001). Feedback inhibition is also critically implicated in hippocampal network oscillations and sparse coding that support hippocampus-dependent learning and memory (Jonas and Lisman, 2014; Colgin, 2016). Therefore, the enhanced feedback inhibition after TBI described here could also be mechanistically linked to cognitive dysfunction that develops after TBI.
In conclusion, the present results show that TBI causes persistent, large, and specific long-term changes to feedback inhibition in the dentate gyrus, particularly involving synaptic output of DGCs born just before the injury. It is unclear whether these altered circuit dynamics represent compensatory changes to address reduced inhibition after TBI or they contribute to seizure susceptibility and behavioral comorbidities of PTE. This study highlights changes in the development and maintenance of immature DGC synaptic input onto PVBCs that may present future opportunities for developing novel treatments for TBI by targeting birth date-specific subsets of DGCs.
Footnotes
This work was supported by R01 NS092552 to B.N.S. The sponsors had no role in study design, data collection, analysis and interpretation, or writing of this manuscript. We thank Dr. Katalin Smith for help with tissue processing; and Tanya Seward for animal maintenance.
The authors declare no competing financial interests.
- Correspondence should be addressed to Bret N. Smith at bret.smith{at}colostate.edu