Introduction

The central nervous system is a plastic structure which may reorganize as a consequence of naturally occurring events, such as learning and memory, but also in response to injury. It is now well established that central nervous system lesions lead to functional loss, but that remaining cells may display plastic adaptations that contribute to the recovery of the function (Nudo et al. 1996; Carmichael and Chesselet 2002; Zepeda et al. 2004; Brown et al. 2007; Winship and Murphy 2008; Sigler et al. 2009). However, reinstatement of the original morphology of the damaged structure does not necessarily accompany such recovery (Zepeda et al. 2004). Recently in view of the continuous generation of new neurons in the adult mammalian subventricular zone (SVZ) (Doetsch and Alvarez-Buylla 1996) and dentate gyrus of the hippocampus (DG) (Kuhn et al. 1996), the process of neurogenesis has gained attention as a possible mechanism for neurorepair. Thus, the DG represents a unique structure to evaluate the impact of plastic events on functional and morphological reorganization after injury given its cellular proliferative potential.

Several works have shown the increased production of endogenous progenitors in the SVZ and DG and the migration of new proliferating cells toward the lesion area in ischemia (Jin et al. 2001; Arvidsson et al. 2002; Parent et al. 2002; Bendel et al. 2005) and epilepsy models (Bengzon et al. 1997; Gray and Sundstrom 1998; Nakagawa et al. 2000; Scharfman et al. 2000). However, naturally occurring structural and functional reorganization is still under debate (for a review see Zepeda et al. 2009).

Stimulation of adult neural progenitors leading to hippocampal morphological and functional reorganization has been achieved after infusion of growth factors (Nakatomi et al. 2002). However, regeneration of injured brain regions where proliferation occurs at a very low rate has not been shown to occur beyond doubt (for review see Okano and Sawamoto 2008). Plasticity in the hippocampus has been mainly shown to occur only when particular combinations of trophic factors are exogenously provided (Nakatomi et al. 2002) although these results have been challenged (Bendel et al. 2005; Sun et al. 2007; Ogita et al. 2005). Thus, it remains controversial whether naturally occurring neurorepair events (i.e. non-stimulated) may be sufficient to account for the morphological restoration of a damaged structure. Although few studies have shown that the DG tends to reorganize morphologically after damage in young (Dong et al. 2003) and adult rodents (Ogita et al. 2005; Hernandez-Ortega et al. 2007), the cellular mechanisms leading to such restoration and its time-dependent functional significance remain unknown.

Given that the DG constitutes a highly plastic structure which is clearly associated with cognitive functions such as the contextual fear memory, we evaluated the structural, behavioral and the long-term potentiation (LTP) of the DG after an excitotoxic focal lesion induced with kainic acid (KA) and addressed the time-dependent correlation between naturally occurring neuronal proliferation and maturation within the structure with functional recovery.

Our results show a time-dependent morphological reorganization of the DG which correlates with the restoration of its volume and is accompanied by modifications in histochemical markers associated to different stages of neuronal maturation. Moreover, induction of in vivo LTP in the DG-CA3 pathway along with functional recovery of DG dependent behavioral tasks was observed in a time-dependent manner.

Materials and methods

Animals

Adult male Wistar rats (n = 54), weighing between 250 and 300 g (3–4 months old) were used throughout the study and handled in accordance with local government rules and the Society for Neuroscience Guide for the Care and Use of Laboratory Animals with approval of the Animal Care Committee of the Instituto de Investigaciones Biomédicas, UNAM. Efforts were made to minimize animal suffering and to reduce the number of subjects used. The total number of animals per group is shown in Fig. 1.

Fig. 1
figure 1

Schematic representation of experimental procedures. On top of the time line, experimental procedures are depicted, on bottom, days corresponding to each procedure. CFC contextual fear conditioning, cued FC cued fear conditioning

Full size image

During the whole experimental procedure, four rats per cage were housed in a laboratory environment with an inverted 12 h artificial light/dark cycle with free access to water and food.

Lesion procedures

For all the experiments, animals were anesthetized with 2–3 % isoflurane in a mixture of 95 % O2 and 5 % CO2 and placed in a stereotaxic system (Stoelting, Wood Dale, Ill). The dentate gyrus (DG) of the hippocampus was localized according to bregma (Paxinos and Watson 1986): AP-3.8, LM-2.4, and DV-3.5. The excitotoxic lesion was induced by a unilateral injection of 1 μl (0.75 nM) of kainic acid (KA; Sigma-Aldrich, Chemie, St. Louis, MO) at a rate of 1 μl/min in the right DG. The KA was dissolved in 1 M NaOH pH 7.0–7.4, and the solution was brought to the desired volume with 10 mM phosphate buffer (pH 7.4). KA was infused using a micro syringe mounted on a microinjection pump (Stoelting, Co., WoodDale, IL, USA). Sham animals received 1 μl/min of 10 mM phosphate buffer. The skin was then sutured, anesthesia was discontinued, and animals were returned to their acrylic cages until the time of perfusion.

During the first hours after the experiment and for the days that followed, the rats were observed until the day of perfusion. None of the rats presented seizures at any time point and only scarce wet-rat shakes were observed in some rats during the first hour after KA infusion.

Apparatus

Open field arena

For the open field motor activity task, an 80 cm long × 80 cm wide × 30 cm high black floor and wall acrylic arena was used. The arena was divided in 20 cm × 20 cm squares for quantifiable locomotion recording.

Place recognition arena and objects

A 60 cm long × 60 cm wide × 30 cm height black floor and wall acrylic arena was used. Objects used for this task were plastic and glass boxes, lids and containers; all were 6–7 cm high, 4–10 cm wide and 4–10 cm deep.

Contextual fear-conditioning chamber

For the contextual fear-conditioning task, a conditioning chamber 25 cm long × 25 cm wide × 20 cm high was used (San Diego Instruments, San Diego, CA, USA). The walls and the roof consisted of transparent acrylic and the removable floor consisted of 23 stainless steel rods. The chamber was equipped with a matrix of 32 infrared beams at the level of the floor. Movements inside the chamber were registered by the interruption of a beam and the information was sent to a computer. The roof of the chamber was equipped with a speaker producing an 80 dB and 2 kHz sound.

Altered context chamber

The fear-conditioning chamber was employed with radical context modifications: the floor was covered with a white acrylic platform, the walls were covered on the outside with a black–white stripe pattern, an acrylic insert was vertically placed inside the chamber so it was divided in two triangular compartments, and a cherry scented ball of cotton was placed below the acrylic floor.

Behavioral testing

A schematic representation of the time line of experimental procedures is shown in Fig. 1.

Animals were tested during the dark cycle, the training and test room was dark and no visual cues were available. At day 4 (n = 15; KA = 7, sham = 8), 19 (n = 16; KA = 8, sham = 8) or 54 (n = 15; KA = 7, sham = 8) after the lesion and for three consecutive days, animals were habituated to the experimenter and to the open field arena. For the next 3 days, animals were trained and tested as follows: on day 7, 22 or 57 subjects were evaluated in a place recognition task; on day 8, 23 or 58, motor behavior in the open field test was assessed and subjects were subsequently trained in contextual fear conditioning and on day 9, 24 or 59 postlesion, we evaluated contextual fear memory and cued fear memory in an altered context.

Behavioral procedures

Open field testing

To evaluate general motor performance animals were individually placed in one corner of the open field arena as previously described, and the arena was divided into 16 squares marked by white lines. Number of crossings was scored for 5 min.

Place recognition test

Place recognition is a spatial memory task particularly modulated by the CA1 of the hippocampus (Ennaceur et al. 1997; Brun et al. 2002; Mumby et al. 2002). Subjects were individually placed in the open field arena with two identical objects located in two opposite corners of the box at 25 cm distance from each corner. Subjects were allowed to freely explore the objects for 5 min (Acquisition period “A”) and then returned to their homecage for 15 min (this period constituted the retention interval), and the time (s) that the subject spent exploring each object was recorded. In the mean time, the objects and the arena were wiped with cleaning solution: 10 % EtOH, 10 % Extran (Merck, Darmstadt, Germany) diluted in water. After this period, one of the objects was located in a new position (novel location) while the other remained at the same location (familiar location). The subject was then allowed to explore both objects for 5 min (Recall period “R”). The session was recorded and exploring time was registered. A recognition index was obtained after calculating: time spent exploring the novel location/(time spent exploring the novel location + time spent exploring the familiar location).

Contextual fear memory

Contextual memories are modulated mainly by the hippocampal DG (Phillips and LeDoux 1992; Lee and Kesner 2004; Saxe et al. 2006). Briefly, a tone (conditioned stimulus) is paired to an electric shock (unconditioned stimulus) in a given context. As a consequence of the conditioning process, subjects develop a conditioned response of freezing, which reflects aversion and is defined as “total absence of movement except for those generated by breathing” (Phillips and LeDoux 1992). The task was performed in the fear-conditioning chamber previously described in the Apparati section. The task consisted of one training and one evaluation session separated by 24 h. Eight, 23 or 58 days after the lesion, subjects were individually trained by placing them in the conditioning chamber where they were allowed to explore for 2 min (pretone) before the first pairing of tone–shock was delivered. The tone had a 20 s duration and co-terminated with a 2 s, and 1 mA foot-shock. Movement was recorded for the following 4 min and four consecutive tone–shock pairings were administered. The total duration of the training session was 22 min. 24 h later, subjects were evaluated for aversive contextual memory: animals were placed individually in the training context and freezing was scored for the entire 5 min session. Subjects were then returned to their home cages. After each session, the chamber was wiped with cleaning solution.

Cued fear memory

Two hours after evaluation in the aversive context, subjects were evaluated for cued aversive memory, which mainly depends on the basolateral amygdala (LeDoux et al. 1990). Each subject was placed in the altered context chamber for 400 s; the tone (20 s duration) used in the contextual fear-conditioning protocol was presented twice, 120 and 260 s after the session had started. No shock was administered and freezing was scored for the next 3 min.

Electrophysiological recordings

Mossy fiber (MF) pathway long-term potentiation (LTP)

After the last behavioral test, animals were prepared for electrophysiological recording for in vivo LTP. Electrophysiological recordings at the MF pathway were performed in anesthetized animals as previously described (Escobar et al. 1997, 2003; Gomez-Palacio-Schjetnan and Escobar 2008, 2010). Responses were recorded using a monopolar microinfusion electrode placed in the CA3 pyramidal layer of the dorsal hippocampus at the following stereotaxic coordinates: AP −2.9 mm, ML +2.2 mm, and DV −3.0 mm (Paxinos and Watson 1986). Responses were evoked using a bipolar electrode via direct and unilateral stimulation of the MF at the coordinates AP −3.5 mm, ML −2.0 mm, and DV −3.1 mm (Paxinos and Watson 1986). Low frequency responses were evoked once every 20 s throughout a 20 min baseline period, after which high frequency stimulation (HFS) (3 trains/100 Hz/1 s/20 s intertrain) was applied and evoked responses were collected for 3 h (Escobar et al. 1997; Schjetnan and Escobar 2010). Mossy fiber LTP was calculated as a percentage of change compared to the mean slope of the field potentials of the entire baseline period prior to the delivery of HFS.

Mossy fiber synapses express presynaptic inhibitory group II of metabotropic glutamate receptors (mGluRs). Thus, a criterion that allows to distinguish MF EPSPs from non-MF EPSPs as well as from area CA3 population spikes, is the selective reduction of this synaptic response caused by the metabotropic glutamate receptor II agonist, 2-(2,3-dicarboxy-cyclopropyl) glycine (DCG-IV).

Microinfusion of DCG-IV (0.5 μl of 5 μM in 5 min) was used at the end of the electrophysiological recordings to verify that the signal was generated by mossy fiber inputs (Calixto et al. 2003). DCG-IV was delivered via microinfusion electrodes which consisted on 33-gauge stainless steel epoxylite-covered cannulae coupled to polyethylene tubing through a 26-gauge stainless steel ring, connected to a Teflon-coated stainless steel wire (0.005-in. diameter). Polyethylene tubing was attached to a 10-μl Hamilton syringe that was driven by a microinfusion pump (Cole Palmer Co. Vernon Hills, IL).

Histological procedures

All chemicals were purchased from Baker (Austin, Tx) unless otherwise stated. After 1, 10, 25 and 60 days of surgery, subjects were anesthetized with sodium pentobarbital and transcardially perfused with 250 ml of 0.9 % chilled saline followed by 250 ml of 4 % chilled formaldehyde in 0.1 M phosphate buffer (pH 7.4). A group of three to five subjects per time point was evaluated for histological analysis. Brains were removed and left in fixative for an additional 24 h. They were successively transferred to 15 and 30 % sucrose, and 30 μm coronal sections were cut in a cryostat (Microm HM550, Thermo Scientific, Waltham, MA, USA); one of each five was mounted on gelatin covered slides and stained with cresyl violet. The rest of the sections were collected in 24-well plates (Corning, NY, USA) filled with cryoprotection solution (25 % ethylene glycol, 25 % glycerol, 50 % PB 0.2 M pH7.4) and stored at 4 °C until ready to be processed for immunohistochemical procedures.

Immunohistochemistry

Free floating sections were processed for brain lipid binding protein (BLBP), calretinin (CR), doublecortin (DCX), neuronal antigen (NeuN), glial fibrillary acidic protein (GFAP) and proliferating cell nuclear antigen (PCNA). Table 1 describes the details of utilized antibodies.

Table 1 Primary antibodies used for immunohistochemical procedures
Full size table

Sections were washed three times for 10 min in PBS–Triton (0.3 %). All sections were then incubated at room temperature for 2 h in 5 % horse serum (1:25; Vector Laboratories, Burlingame, CA) diluted in 0.3 % Triton-X-100 in PBS, except for those treated for PCNA where the antigen was first recovered at 80 °C for 20 min in 10 mM sodium citrate solution, pH 6.0. After blocking, sections were incubated overnight at 4 °C in anti NeuN or in combination of two primary antibodies, BLBP/PCNA or DCX/CR or DCX/NeuN or GFAP/NeuN (see Table 1 for details) diluted in blocking solution. Sections were washed three times for 10 min with PBS and then incubated for 1 h at room temperature with the corresponding secondary antibodies (see Table 2) diluted in blocking solution. In some cases, sections were counterstained with DAPI (1 μg/ml) (Invitrogen, Carlsbad, CA). Sections were mounted in Superfrost glass slides (Thermo Scientific, Waltham; MA) with the antifading medium; 1,4-diazabicyclo octane solution (DABCO) (Sigma, St. Louis, MO).

Table 2 Secondary antibodies used for immunohistochemical procedures
Full size table

Confocal microscopy and image analysis

Histological sections where cell counting was performed were captured and analyzed using the Zeiss LSM5 confocal microscope (Oberkochen, Germany). Cell counting was performed on three to four 20 μm sections from three animals for each experimental condition. Each section analyzed was separated by 40 μm from the next and contained information from a 14 μm optical section in the axial plane. Four fields per section at a 40× magnification within the granular layer were randomly analyzed for the different cell markers. Sections in which counting was performed corresponded to the injured region (as observed by the disorganized granular layer) and were contained within coordinates AP −3.14 and −4.30 mm according to Paxinos and Watson (1986). Images were exported to Image J software (version 1.42q, Wayne Rasband, National Institutes of Health); the number of positive cells per optical field was recorded and average was calculated. A student′s t test was performed to validate significant differences between sham and lesion groups.

Volumetric analysis of the Dentate Gyrus

Dentate granule cell layer volume of Nissl stained and NeuN immunohistochemically processed sections was calculated using a stereological method, the Cavalieri Estimator Protocol from StereoInvestigator software (MBF Bioscience, Inc., Williston, VT). 12 sections per animal each separated by 120 μm and contained within AP −3.2 and −4.8 were randomly evaluated. The contour of the DG from all sections was delineated, the volume of the granular layer from each subject was automatically calculated and the average per group (10, 25, or 60 day postlesion; sham and KA) was obtained and reported in mm3. Contours were exported to Neurolucida software (MBF Bioscience Inc., Williston, VT) and reconstructions from the DG were produced to create a visual model of the volume of the structure.

Results

Excitotoxic focal lesion in the dentate gyrus in time

The evidence of the lesion as assessed through Nissl staining is shown in Fig. 2. Coronal sections show a well-delimited region in the DG where granular cells were absent or where the morphology of the DG in animals killed 10 days postlesion (dpl) was altered as compared to the sham group (Fig. 2a, b); this observation was true for 6/7 subjects. At 25 dpl, the DG dorsal layer appeared continuous, but with slight deformations (Fig. 2c) and in 7/7 animals killed at 60 dpl the DG appeared completely reorganized (i.e. the lesion site was not evident) throughout the anteroposterior and dorsoventral axes (Fig. 2d). Well-delimited and sparse neuronal death was also observed in the CA1 and CA3 subfields, respectively; damage was still evident in these regions until 25 and 60 dpl corroborating the selectivity of DG morphological reorganization (Fig. 2b–d).

Fig. 2
figure 2

Reconstruction of Nissl stained dentate gyrus coronal sections captured at ×10 were assembled to reconstruct the DG. a sham, b 10 dpl, c 25 dpl, d 60 dpl. Insets show a ×40 magnification of the region shown in black frames. A’–D’ correspond to the crest of the DG and A’’–D’’ correspond to the dorsal layer of the DG

Full size image

A volumetric analysis using the Cavalieri probe (MBF, Bioscience Inc., Williston, VT) for measuring the DG (Fig. 3a) showed that the average granular layer volume in the 10 dpl group was significantly diminished as compared to the sham, 25 and 60 dpl groups (F (3,15) = 11.232; p < 0.01, one way ANOVA). However, the DG volume from the 25 dpl group did not differ from the sham or 60 dpl groups (Fig. 3b). All the Cavalieri analyses yielded a Gundersen coefficient <0.05 (Table 3) (Gundersen and Jensen 1987). These results show that the granular layer recovers its original volume with time after the lesion.

Fig. 3
figure 3

Visual modeling and volumetric analysis of the dentate gyrus. a Neurolucida projection from a Nissl stained coronal section showing the DG of a sham animal, b stereological analysis of the granular layer of the DG reveals that the volume in the 10 dpl group significantly differs from sham, 25 and 60 dpl (F (3,15) = 11.232; **p < 0.01)

Full size image
Table 3 Volume and Gundersen′s coefficient corresponding to the average per group of the granular layer of the DG
Full size table

Morphological reorganization of the DG

To analyze cellular proliferation and neuronal maturation events associated to the morphological reorganization of the DG, a multiple immunofluorescence and confocal microscopy analysis for DCX- and CR-positive neurons and NeuN-positive nuclei was performed in brains from animals killed at 1, 10 and 60 dpl. DCX is a microtubule marker widely used to evaluate neurogenesis levels, as its expression occurs in neuronal committed newly born cells (Rao and Shetty 2004; Couillard-Despres et al. 2005), whereas CR expression relates to an early postmitotic step of neuronal differentiation (Kempermann et al. 2003). The DG of animals that received a KA injection and were killed at 1 dpl, showed a twofold increase in DCX expression as compared to sham subjects (p < 0.05; t test). Such increase progressed reaching the highest level both in the ipsi and contralateral sides to the lesion at 10 dpl. A student’s t test showed significant differences between 10 dpl and sham groups (p < 0.05) (Fig. 4a–c and right middle panel). After 60 dpl, when the DG appeared completely restructured, the number of DCX-positive cells was similar to that observed in the hippocampus from sham animals (Fig. 4d and right middle panel). In sham animals, DCX-positive cells exhibited the phenotype of undifferentiated granule cells with short dendrites mostly oriented horizontally toward the SGZ (Fig. 4e). In addition to the rise in DCX-positive cell number at 1 dpl, a more differentiated phenotype was observed and consisted in an increased number of elongated dendrites vertically oriented toward the granular cell layer (Fig. 4f) which contrasted with the short and horizontally oriented neurites found in sham injected animals. At 10 dpl, there were abundant DCX-expressing dense cell clusters in the SGZ (Fig. 4g), suggesting that an intense process of neuronal proliferation was taking place. However, at 60 dpl, the total number and the appearance of DCX-positive cells returned to control levels (Fig. 4h and right middle panel). Although KA was applied unilaterally, an increase in markers of cell proliferation was also observed in the intact hemisphere. It is thus conceivable that the neuronal excitation elicited by KA might propagate to the contralateral hippocampus through commissural fibers and transiently increase cell proliferation without producing structural changes (Fig. 4, bottom panel). In this sense, it has been reported that the unilateral administration of neurotoxic compounds leads to the expression of stress proteins in the intact hemisphere (Arias et al. 1998; Ayala and Tapia 2008).

Fig. 4
figure 4

Immunohistochemistry for doublecortin (DCX) and calretinin (CR) in coronal sections from an injured DG. Blue, cell nuclei stained with DAPI. DCX (green) and CalR (red) (a–d). Representative sections from a sham lesion animal (a); and subjects killed at 1 dpl (b); 10 dpl (c) and; 60 dpl (d). e–h Higher magnifications of different regions of the DG from a sham animal and subjects killed at 1 dpl (f), 10 dpl (g) and 60 dpl (h); DCX (green), NeuN (red). Arrows point at morphological differences between DCX-positive cells from a sham subject (e) and another killed 1 dpl (f); arrowhead indicates the appearance of DCX-expressing dense cell clusters at 10 dpl (g). Photomicrographs are representative of 3–4 sections from 3–5 different animals of each experimental group. Values represent mean ± S.E.M. of three different experiments where DCX-positive cells (right middle panel) and CalR-positive cells (right bottom panel) were counted; I ipsilateral, C contralateral. Asterisks denote significant differences as compared to sham (student’s t test) p < 0.05. Scale bars 50 µm

Full size image

The inner molecular and infragranular layers of the DG displayed CR immunoreactivity in sham animals (Fig. 4a). A three-fold increase in CR-positive cells was observed as soon as 1 dpl and until 60 dpl in the SGZ as well as in the hilus (Fig. 4b–d and right bottom panel). Interestingly, after 10 and 60 dpl, the CR immunoreactivity was also strongly observed in the inner molecular and infragranular layers of DG. Increase in CR expression was significantly different from the sham group at all time points in the ipsilateral DG, but also in the contralateral DG at 10 dpl (p < 0.05; t test).

To determine changes in the contents of neural progenitor cells, we performed a quantitative analysis of BLBP-positive cells in view that this protein is considered as a specific marker of radial glia-like cells in the SGZ (Steiner et al. 2006). At 1 dpl, few BLBP/PCNA-positive co-localizing cells were observed (Fig. 5b). The content of BLBP-positive cells significantly increased at 10 dpl in the damaged DG (including the SGZ and the hilus) as compared to the sham group (p < 0.05; t test) (Fig. 5a–c and right middle panel), and at 60 dpl, labeling of this marker returned to basal levels (Fig. 5d).

Fig. 5
figure 5

BLBP-positive glia-like neuroprecursor cells in DG at different time points after KA lesion. Coronal sections showing a triple immunofluorescence of BLBP cells (red), DAPI-labeled cell nuclei (blue) and proliferating cells with PCNA (green) (a–d). At higher magnifications of regions shown in a–c, BLBP-positive cells with PCNA nuclei were observed at 1 dpl (f, arrowhead), but not at 10 or 60 dpl (g–h). At 10 dpl a robust increase of BLBP was evident in the subgranular zone and in the hilar region (g, arrows). Quantification of cell number positive to BLBP and PCNA in the injected (I) and contralateral hippocampus (C) is shown. Photomicrographs are representative of 3–4 sections from 3–5 different animals of each experimental group. Values represent mean ± S.E.M. of three different experiments. Asterisks denote significant differences as compared to sham (student’s t test) p < 0.05. Scale bars 50 µm

Full size image

The proliferation cell nuclear antigen (PCNA) showed an increase from day 1 to 60 postlesion as compared to sham levels (p < 0.05, t test) (Fig. 5a–d) reaching the highest point at 10 dpl (Fig. 5c and right bottom panel). However, colocalization with BLBP was mostly evident at 1 dpl (Fig. 5e–h).

To evaluate the presence of the glial response to damage, we analyzed GFAP-positive cells. Figure 6 shows the distribution of radial glia-like cells in the sham DG. Loss of GFAP-positive cells became evident as soon as 1 dpl in the SGZ (compare Fig. 6A, a with B,b) and processes in remaining cells shortened and diminished in number. However, at 10 dpl radial glia-like cells became distinguishable again (compare Fig. 6a with c). Remarkably, the GFAP-positive astrocytic response typical of damage was confined to the hilus but was not observed within the injured SGZ region (Fig. 6B, C).

Fig. 6
figure 6

GFAP positive cells at 1 and 10 d after KA lesion. Coronal sections labeled with NeuN (red) and GFAP (green) of sham injected animals show the presence of radial-like glial cells (A and a). At 1 dpl, GFAP-positive radial processes shortened and diminished in number (arrow, compare a and b) and at this time point as well as at 10 dpl a gliotic reaction was observed in the injured dorsal blade of DG without entering into the granular cell layer (B, C arrowheads). At 10 dpl some radial-like glial cells became evident (arrow, c). Photomicrographs are representative of 3–4 sections from 3–5 different animals per group. Scale bars 40 µm

Full size image

To further analyze the cellular contents of the reorganized region, we performed a NeuN immunohistochemistry at all time points. Interestingly the DG of 6/7 animals killed at 10 dpl showed an area devoid of NeuN-positive nuclei (compare Fig. 7a, b) within the granular layer, whereas at 25 and 60 dpl none of the analyzed sections showed such a region, instead NeuN-positive cells were uniformly distributed along the DG in the whole AP axis (Fig. 7c, d). Moreover, the hilar region appeared devoid of NeuN-positive cells at 10 and 25 dpl, while scarce cells were observed at 60 dpl. Thus, structural reorganization at the later time points studied seems to be particular of the granular layer of the DG.

Fig. 7
figure 7

NeuN immunohistochemistry reveals the morphological reorganization of the DG in time. Six to seven images captured at ×10 were assembled to reconstruct the DG. a Sham; b 10 dpl; c 25 dpl; d 60 dpl; scale bar 400 μm. Insets show a ×40 magnification of the region shown in white frames. A’, B’, C’ and D’ correspond to the dorsal layer of the DG and A’’, B’’, C’’ and D’’ correspond to the crest of the DG; scale bar 100 μm

Full size image

Focal dentate gyrus lesion impairs contextual fear memory, but functional recovery occurs in time

To address the functional impact of the lesion on behavior, we evaluated all groups in the cued fear memory and contextual fear memory tasks. Cued memory is known to rely on the amygdala, whereas contextual fear memory depends on the hippocampus and particularly on the DG (Phillips and LeDoux 1992). Intact, sham and lesioned animals received five tone–shock pairings in a single session. Figure 8a shows that in the 2 min lapse before the tone–shock pairing, subjects displayed a normal movement behavior, which translates into low percentage of freezing. However, freezing increased gradually after each pairing in all groups, thus animals were successfully conditioned regardless of the time postlesion when they were trained (Fig. 8b). No significant differences in freezing were obtained among groups (F (6,47) = 1.309, p = 0.1624, repeated measures ANOVA). 24 h after the training session, animals were evaluated for contextual fear memory and, 2 h later, in cued fear memory in an altered context. Figure 8c shows that cued fear memory was preserved in all lesion groups (i.e. percentage of freezing time was not different from sham groups at any time point) (F (6,47) = 0.827, p = 0.55, factorial ANOVA). These results differed from those observed after evaluating contextual fear memory. Figure 8d shows that only the 10 dpl group spent significantly less time freezing than its control group in the aversive context (F (6,47) = 13.587, p < 0.0001, factorial ANOVA) whereas freezing time in such context in the 25 and 60 dpl groups was not significantly different from the sham or intact groups, thus revealing an initial loss and a subsequent functional recovery of the impaired behavior.

Fig. 8
figure 8

Contextual fear-conditioning recovery in time. a Freezing percent of time spent during the pretone period (120 s). All groups displayed a similar low percentage (F (6,47) = 1.312, p = 0.27; factorial ANOVA) of freezing in the conditioning context before the tone–shock pairings were administered. b Percentage of time spent freezing during the conditioning session. All groups increase the freezing time according to the presentation of tone–shock pairings; “n” per group is depicted in parenthesis. (F (6,47) = 1.309, p = 0.1624, Repeated measures ANOVA). c Evaluation of freezing in the cued fear memory task, all groups display an elevated percentage of time spent freezing which did not differ among each other (F (6,47) = 0.827, p = 0.55; factorial ANOVA). d Freezing behavior in the contextual fear memory task. Significant differences were observed between 10 dpl and the rest of the groups (F (6,47) = 13.587, **p < 0.0001; factorial ANOVA)

Full size image

Given that freezing behavior implies lack of movement, we tested the animals in the open field test to guarantee that freezing was not caused by motor alterations or by the lack of interest in exploring the environment. Figure 9a shows no significant differences in the number of visited squares in the open field test among groups at any time point evaluated thereby confirming that freezing was dependent on the conditioning and not on motor impairment or emotional distress.

Fig. 9
figure 9

Behavior in control tasks independent from the DG. a Performance in the open field shows motor facilitation in all subjects without group differences (F (6,47) = 1.561, p = 0.1797; repeated measures ANOVA). b Performance in a place recognition task shows a similar exploratory index among groups between the familiar and novel location. A Acquisition period, R recall period (F (6,47) = 0.304, p = 0.9316, repeated measures ANOVA)

Full size image

In addition, we explored the functional outcome of the lesion in the place recognition task which has been shown to be particularly modulated by the CA1 subfield of the hippocampus. Figure 9b shows that place memory was not impaired by the DG lesion at any time point evaluated and that injured animals spent a similar amount of time exploring the new location of a known object as the sham and intact groups (F (6,47) = 0.304, p = 0.9316, repeated measures ANOVA). Thus, the deficit provoked by the lesion was restricted to a function mainly modulated by the DG.

Long-term potentiation recovers in time

To evaluate whether behavioral recovery and structural reorganization correlated with synaptic plasticity of the DG, we induced high frequency LTP in vivo on the mossy fiber pathway (DG-CA3). Figure 10a (top left panel) shows a schematic representation of the mossy fiber pathway and electrode placement. The CA3 EPSP consisted of potentials of 0.38 ± 0.07 mV (mean ± SEM) elicited with 20–40 μA current pulses of 0.1–0.25 ms duration. These responses began at 1.8–3.3 ms poststimulation and presented their valley at 7.1 ms with an average slope of 0.15 ± 0.007 (mean ± SEM) in agreement with previous studies (Escobar et al. 1997; Gomez-Palacio-Schjetnan and Escobar 2008; Schjetnan and Escobar 2010). Microinfusion of DCG-IV at the end of the electrophysiological recordings selectively blocked mossy fiber responses to 15 % of baseline signals, presenting potentials of 0.09 ± 0.02 mV with a mean slope of 0.06 ± 0.001 (mean ± S.E.M.), verifying that the signal was generated by mossy fiber inputs (Fig. 10a, right panel).

Fig. 10
figure 10

Mossy fiber LTP induction in vivo by HFS recovers in time. a Left schematic representation of the mossy fiber pathway and electrode placement showing the stimulated (CA3) (a) and recorded (DG) (b) sites in a coronal plane. Right representative traces of the EPSP obtained at baseline (full line) and after DCG-IV application (dotted line). Note that DCG-IV selectively blocked mossy fiber responses to 15 % of baseline signals. Plot of mossy fiber evoked responses from the intact group (B); 10 dpl (C); 10 d sham lesion (D); 25 dpl (E); 25 d sham lesion (F); 60 dpl (G); 60 d sham lesion (H). DG-CA3-LTP induction was only blocked in the 10 dpl group. Inner numbers show mean ± S.E.M. EPSPs slope (percent of baseline) obtained 1 h after HFS for each group. The trace on the top left in each graph shows the average of MF field potentials obtained immediately before (full line) and 170 min after (dotted line) HFS

Full size image

The present results show that high frequency stimulation (HFS) (three trains 100 Hz/1 s/20 s intertrain) was capable of inducing LTP at the DG-CA3 pathway in adult rats in vivo in accordance with previous studies (Escobar et al. 1997; Gomez-Palacio-Schjetnan and Escobar 2008; Schjetnan and Escobar 2010). Mossy fiber LTP was characterized by a slow initial increase in the EPSP slope that has been related to the independence of NMDA receptors activation and opioid peptide relevant modulation (Derrick et al. 1991, 1992; Escobar et al. 1997). At 1 h post-stimulation, the intact group presented a mean EPSP increase (mean ± SEM) of 147.71 ± 1.45 (Fig. 10b). However, at 10 dpl DG-CA3-LTP could not be elicited (Fig. 10c), as opposed to the 10-day sham lesion group (Fig. 10d). Repeated–measures ANOVA for slope increases revealed significant differences among groups (F 6,23 = 7.15) and a post hoc Fisher’s test showed significant differences between the 10 dpl group and all of the other groups (p < 0.001). Interestingly, at 25 and 60 dpl, LTP could be elicited and did not differ from the corresponding sham or intact groups (Fig. 10e–h). HFS-induced potentiation was not associated with after discharges or other overt epileptiform activity as assessed by the continuously recorded hippocampal EEG (data not shown).

Discussion

In this study, we show that the DG is capable of reorganizing functionally and structurally in time after undergoing a unilateral focal lesion and that reorganization relied exclusively on endogenous mechanisms. We demonstrate that the granular cell layer of DG is able to repopulate after a focal lesion and that the structural reorganization correlated in time with the recovery of contextual memory function and with in vivo restoration of LTP in the DG-CA3 pathway.

It is feasible that newly added granule cells played a role in this remodeling process, as we detected a long-lasting increase in hippocampal neurogenic markers corresponding in time with the refill of the lesion site. We evaluated several markers to make a follow up of the morphological reorganization process. Cells in the SGZ expressing BLBP are undifferentiated, radial glia-like neural stem cells (Suh et al. 2007). As early as 24 h after the lesion, few BLBP-positive cells with PCNA-positive nuclei were already present reaching the highest levels at 10 dpl. Thus, the beginning of proliferation of these precursors started soon after the lesion. PCNA remained high at all time points as compared to sham levels reflecting proliferation of several kinds of cells, but also likely suggesting an ongoing process of DNA repair (Eisch and Mandyam 2006).

As soon as 24 h after KA injection, we also detected the increase in DCX-positive cells with vertically oriented cellular processes as opposed to the short and horizontally oriented neurites found in sections from sham animals. This occurred prior to a robust rise in cells expressing BLBP in agreement with the idea that most radial glial-like cells are quiescent and only a short subset is recruited to proliferate (Ledergerber et al. 2006). Thus, a group of DCX-expressing cells might contain proliferative type-2 progenitor cells (Jessberger et al. 2005; Hodge et al. 2008) which may start a neurorepair process. It has been shown that DCX-expressing cells within the DG reemerge after traumatic brain injury and it has been suggested that they are the more likely contributors to stable neurogenesis (Yu et al. 2008; Kernie and Parent 2010). Although similar changes in number and morphology of DCX-positive cells have been reported after systemic KA-induced seizures (Gu et al. 2010) we cannot attribute this observation to seizures because we used a subepileptogenic dose and only few rats showed isolated wet-rat shakes during the first hour after infusion.

It was noteworthy that the number of CR-positive cells increased as early as 24 h after KA lesion. CR has been considered as a marker of interneurons and Cajal–Retzius cells in the DG, which co-express the glycoprotein reelin (Tissir and Goffinet 2003) a pivotal factor in maintaining the lamination in the hippocampus as well as in the rest of the brain (Duveau et al. 2011). Interestingly, CR immunoreactivity was increased also at the sharp borders toward the outer molecular layer of the DG as reported in studies of morphological organization of this structure (Del Turco et al. 2004). In cortical lesion models, newly generated cells may differentiate into neurons and glia in the regenerating cortex, but lamination is not reinstated (Kolb et al. 2007).

Our results show that at 60 dpl, the zone originally devoid of cells was refilled with NeuN-positive nuclei. Thus, even when we cannot rule out the possibility that plastic events involving neurite reorganization of preexisting neurons is contributing to the observed recovery (Zepeda et al. 2004) the replenishing of the granule cell layer is associated in time with functional reorganization. However, further studies are required for establishing the precise adaptive mechanisms underlying DG restoration.

Our results show an increase of neurogenic markers also in the contralateral hippocampus, which are not necessarily associated to the morphological reorganization process in the damaged granular layer, but with molecular and cellular signals which may activate the contralateral hippocampus through commissural fibers. The cellular response of a lesion in the contralateral region of an injured structure has been previously documented (Arias et al. 1998; Ayala and Tapia 2008).

Functional recovery has been extensively shown to occur after brain injury (for review see (Murphy and Corbett 2009). However, structural disorganization may prevail despite behavioral recovery (Kolb et al. 2007).

Our results showed that 60 days after a focal lesion to the DG it reorganizes anatomically to the point that (1) the volume of the structure recovers to control levels and, (2) the lesion site is no longer evident as revealed by NeuN immunohistochemistry. Interestingly, such structural reorganization has not been documented after cortical damage and we did not detect at any time point the glial scar typical of cortical lesions (Zepeda et al. 2003, 2004). Both, the core and the scar have been proposed to form a restrictive environment for neural repair processes to occur (see Robel et al. 2011 for review). Thus, the cortical and DG milieus may substantially differ and the latter may provide a more permissive environment for cellular plastic events to occur.

Morphological restoration may in turn give rise to the recovery of the aversive memory function, which importantly relies on the DG integrity (Saxe et al. 2006). We can be certain that the animals formed an aversive association through the learning period, given that freezing increased within the session in all groups. This result is in line with previous studies showing that the dorsal hippocampal neurons are not required for learning context fear, but are required for fear context retrieval (Maren et al. 1997). Thus, the initial failure in freezing reflects a memory loss and not an acquisition deficit, whereas freezing after 25 and 60 dpl reveals the ability to remember the aversive context. Generalized freezing can be discarded given that motor behavior in the open field as well as in the 2 min period prior to the tone appearance in the altered context remained intact (data not shown). A work by Winocur et al. (2006) has previously shown that inhibition of neurogenesis after irradiation provokes a deficit in fear memory for as long as 2 months after ablation, thus suggesting a role of new neurons for the adequate performance on hippocampus-dependent memory function. Also a previous work has shown that after a traumatic brain injury, reappearance of neurons correlate with behavioral recovery and the DG has been proposed as the neurogenic source (Sun et al. 2007). Our results show that reorganization, as evaluated with NeuN, occurs along functional recovery and within 25 days which correlates with the time-window in which newly generated neurons have been shown to mature (Zhao et al. 2006).

In addition to the morphological and behavioral reorganization, the parallel reinstatement of LTP occurred. Our results show that at 25 and 60 dpl, HFS was capable of inducing in vivo LTP at the DG-CA3 pathway that was blocked 10 dpl. These results are supported by previous studies showing that low dose of gamma radiation, drastically reduced new cell proliferation, selectively blocked LTP at DG in vitro (Snyder et al. 2001) and prevented the LTP recovery after ischemic damage (Wang et al. 2005). To our knowledge, mossy fiber LTP after a DG lesion has not been previously addressed. Thus, we can only speculate that the parallel reinstatement of LTP along with the morphological and behavioral reorganization could rely on well described plastic adaptive mechanisms such as sprouting of fibers, and modifications in the excitatory/inhibitory balance, but the idea that new neurons may be participating in the organization of a preexisting neuronal network should also be considered (Kempermann et al. 2002, 2004).

Of particular interest, functional and morphological reorganization in our study occurred in the absence of exogenous molecules known to enhance plastic responses in the CNS (Yoshimura et al. 2001; Kolb et al. 2007). Moreover, the time lapse in which recovery occurred is consistent with the time that takes new neurons to migrate and integrate into the rodent DG circuit (Cameron et al. 1993; Zhao et al. 2006).

In conclusion, our results provide information on the progression of the cellular processes that are triggered after a neurogenic niche injury in the adult and show that the restoration of the original morphology of the damaged structure occurs in parallel to the functional recovery of the DG, as assessed through behavioral and electrophysiological approaches.