Cerebral Cortex Advance Access originally published online on April 13, 2005
Cerebral Cortex 2006 16(1):47-63; doi:10.1093/cercor/bhi083
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Age-dependent Spontaneous Hyperexcitability and Impairment of GABAergic Function in the Hippocampus of Mice Lacking trkB
1 Department of Cell Biology and IRBB-Barcelona Science Park, University of Barcelona, Barcelona E-08028, Spain, 2 Present address: Department of Molecular Biochemistry and Biophysics, Karolinska Institute, 171 77 Stockholm, Sweden
Address correspondence to Dr Fernando Aguado, Barcelona Science Park, University of Barcelona, Modular Building (A1–S1), Josep Samitier, 1–5, Barcelona E-08028, Spain. Email: faguado{at}pcb.ub.es.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Patterned intrinsic network activity plays a central role in shaping immature neuronal networks into functional circuits. However, the long-lasting signals that regulate spontaneous activity of developing circuits have not been identified. Here we study the net impact of TrkB signaling on early network activity of identified neuronal populations by analyzing postnatal hippocampi from trkB null mice. Ca2+ imaging showed that pyramidal neurons of trkB–/– mice displayed a decrease in intrinsic synchronous activity in neonatal animals but an increase in juveniles. Strikingly, alterations in network activity in trkB–/– hippocampus were associated with an aberrant induction of the transcription factor Fos. In contrast to pyramidal neurons, spontaneous [Ca2+]i oscillations in trkB–/– interneurons were consistently impaired throughout postnatal development. Moreover, the number of GABAergic synapses and the expression levels of GAD65 and KCC2 were decreased in mutant hippocampi, indicating that pre- and post-synaptic GABAergic components were impaired in trkB–/– mice. Finally, the partial blockade of GABAA receptor in postnatal slices revealed that mutant hippocampi displayed an increased susceptibility to network hyperexcitability. These results indicate that the lack of TrkB signaling during development impairs GABAergic neurotransmission, thereby leading to an age-dependent decrease followed by an increase in the intrinsic excitability of neuronal circuits. Furthermore, the present study indicates that long-lasting TrkB signaling may contribute to the construction of CNS circuits by modulating patterns of spontaneous [Ca2+]i oscillations.
Key Words: calcium oscillations • circuit maturation • GABAergic interneurons • KCC2 • neurotrophins • seizures • spontaneous activity
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In the cerebral cortex the fine-tuned balance between excitation and inhibition synchronizes discharges of pyramidal neurons to generate a variety of network activities (Freund and Buzsaki, 1996
; Steriade, 1997; Klausberger et al., 2003). The earliest synchronous network activity occurs at the time the first synaptic contacts have been formed to correlate the firing of hundreds of neurons (Yuste et al., 1992; Feller, 1999; Ben-Ari, 2001). This robust correlated activity is spontaneously generated by the regenerative overexcitation of immature microcircuits and progressively decreases as experience-dependent activity emerges (Katz and Shatz, 1996; Ben-Ari, 2002). It is believed that early intrinsic synchronous activity controls neuronal development to shape primitive and simple networks into functional circuits (Katz and Shatz, 1996; Ben-Ari, 2001, 2002; Zhang and Poo, 2001). Therefore, it has been proposed that pathological function of adult circuits, as occurs in schizophrenia and epileptic seizures, is the deleterious consequence of altered intrinsic network activity during development (Mizrahi and Clancy, 2000; Ashe et al., 2001).
Glutamate and electrical coupling generate spontaneous correlated network activity in the developing cerebral cortex (Yuste et al., 1995; Garaschuk et al., 2000; Ben-Ari, 2001). Moreover, intrinsic synchronous and oscillatory firing of cortical microcircuits depends on the pivotal role of GABAergic transmission (Freund and Buzsaki, 1996; Ben-Ari, 2002). For instance, the depolarizing action of GABA triggers network activity in the embryonic and neonatal hippocampus, whereas synchronous output of mature pyramidal neurons is driven by GABAergic inhibition (Freund and Buzsaki, 1996; Ben-Ari, 2002; Klausberger et al., 2003).
In contrast to triggering mechanisms, the identities of long-lasting signals that regulate developing synchronous network activity have not been established. Members of the neurotrophin family are important signals that control neuronal excitability and activity-dependent neuronal plasticity (Schinder and Poo, 2000). In particular, acute and long-lasting activation of the tyrosine kinase receptor TrkB by its ligand BDNF alters neuronal excitability and is required for the induction of LTP in the hippocampus, respectively (Korte et al., 1995; Kafitz et al., 1999; Minichiello et al., 1999). A body of evidence has shown that developmental activation of TrkB controls axonal growth, dendritic arborization and synaptic transmission of cortical glutamatergic neurons (for reviews, see Huang and Reichardt, 2001; McAllister, 2001). Interestingly, more recent studies have demonstrated that maturation of the GABAergic system is a key target for TrkB signalling (Bao et al., 1999; Huang et al., 1999; Seil and Drake-Baumann, 2000; Richardson and Leitch, 2002; Rico et al., 2002; Jin et al., 2003; Elmariah et al., 2004; Palizvan et al., 2004). However, the in vivo role of BDNF/TrkB signaling in controlling the balanced intrinsic activity of developing circuits is poorly understood. Recently, we showed that targeted BDNF overexpression in neural precursor cells increased the onset of spontaneous correlated network activity in the embryonic hippocampus (Aguado et al., 2003). Because of the wide-spectrum actions of BDNF on various aspects of neuronal differentiation and maturation, here we studied the net effect of endogenous TrkB signaling on the intrinsic activity of developing postnatal hippocampus by inactivating the trkB gene. Spatio-temporal patterns of activation were analyzed in particular neuronal populations by Ca2+ imaging (Yuste et al., 1992; Wong and Oakley, 1996; Mao et al., 2001). In the present study we show that the components and function of GABAergic neurotransmission were consistently impaired in mutant mice. Moreover, trkB–/– mice showed age-dependent alterations in spontaneous network activity, which were associated with an aberrant induction of the transcription factor c-fos. Finally, an increased susceptibility to network hyperexcitability was detected in juvenile trkB-deficient hippocampi.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Animals
Heterozygous C57BL/6 breeding pairs carrying a germline mutation in the tyrosine kinase catalytic domain of the trkB gene (trkB+/–) (Klein et al., 1993) were a generous gift from Drs Mariano Barbacid and Inmaculada Silos-Santiago (CNIO-CNB, Madrid, Spain and Millenium Pharmaceuticals, Boston, MA). trkB–/– mice were generated by mating heterozygous animals and genotypes were identified by PCR amplification of tail DNA. The animal colony was kept under controlled temperature (22 ± 2°C), humidity (40–60%) and light (12 h cycles) conditions, and treated in accordance with the European Community Council Directive (86/609/ECC). Mice were analyzed at two postnatal stages: neonatal (P2–3) and juvenile (P8–13). Wild-type littermates were used as controls in all experiments.
Ca2+ Imaging
We studied spontaneous neuronal activity by recording non-evoked [Ca2+]i changes on hippocampal acute slices from P2–3 and P8–9 stages (Yuste et al., 1992; Leinekugel et al., 1997; Garaschuk et al., 1998; Schwartz et al., 1998; Mao et al., 2001; Aguado et al., 2003). Brains were removed and placed in cold artificial cerebro-spinal fluid (ACSF; in mM: NaCl 120, KCl 3, D-glucose 10, NaHCO3 26, NaH2PO4 2.25, CaCl2 2, MgSO4 1, pH 7.4, bubbled with 95% O2/5% CO2). Horizontal brain slices including the hippocampus and the adjacent entorhinal cortex (300 µm thick) were cut with a Vibratome, and before imaging were kept for at least 1 h in a storage chamber containing ACSF bubbled continuously with 95% O2/5% CO2 at room temperature (22–25°C). [Ca2+]i in slices was measured with the membrane-permeant acetoxymethyl ester of fura-2, fura-2-AM (Molecular Probes, Eugene, OR) dissolved in dimethyl sulfoxide with 0.001% pluronic acid (Molecular Probes). The tissue slices were incubated in 3–5 µl of 5 mM fura-2-AM for 2 min and then in 3 ml of 10 µM dye in ACSF for 15 min, as previously described (Schwartz et al., 1998; Badea et al., 2001; Aguado et al., 2003). Slices were always maintained in oxygenated ACSF.
Fura-2-loaded slices were transferred to a continuously superfused recording chamber on the stage of a fluorescent upright microscope (BX50WI; Olympus, Tokyo, Japan) equipped with 380 and 340 nm excitation filters and differential interference contrast (DIC) optics. Recordings of [Ca2+]i changes were imaged with 20x and 40x water-immersion objectives at room temperature (22–25°C). Images were captured with a silicon-intensifier tube camera (Hamamatsu C2400-08) and a frame grabber (LG-3; Scion Corp., Frederick, MD) connected to a Macintosh computer (Apple Computers, Cupertino, CA). Fura-2 fluorescence images were collected at 4 s intervals (15 frames were averaged for each time point) at a single excitation wavelength using the 380 nm bandpass filter over periods up to 15 min controlled by the NIH Image program. To prevent photobleaching, a shutter (UniBlitz, Rochester, NY) controlled by custom-written macros was used. Ethyleneglycol-bis(ß-aminoethyl)-N, N, N, N, tetraacetic acid (EGTA), tetrodotoxin (TTX) and (–)-bicuculline methiodide (BMI) were obtained from Sigma (St Louis, MO).
Network Analysis
To analyze coactive networks of optical recordings from identified cell types, we followed a methodology previously described in detail (Schwartz et al., 1998; Aguiló et al., 1999; Badea et al., 2001; Aguado et al., 2002, 2003). Changes in fluorescence in multiple cells were analyzed with a program written in Interactive Data Language (IDL; Research Systems Inc., Boulder, CO). The fluorescence change over time was defined as 
F/F = (F0 – F1)/(F0). The onset of each Ca2+ transient for every cell was determined using an algorithm that defined the onset as the frame after which the F/F change was larger than a given set threshold, typically a 3–5 pixel value unit change per frame. Ca2+ transients detected by the program were carefully inspected, and spurious events were cancelled (Schwartz et al., 1998). The imaged cells not exhibiting Ca2+ transients over the total recording time (800 s) were considered non-spontaneous active cells. The time of initiation of each Ca2+ transient for each active cell was marked in a raster plot. These plots were used to calculate the matrix of asymmetric correlation coefficients between all cell pairs. Contingency tables were then used and 
2 tests were run to detect the correlation coefficients that were significantly greater than expected. Significant correlation coefficients were used to generate a correlation map on which lines link neurons whose asymmetric correlation coefficient is significant (P < 0.01) and on which the thickness of a line connecting any two cells represents the size of the greater asymmetric correlation coefficient between the cells. To examine the synchronous coactivations that recruited >50% of active pyramidal neurons within the network, we analyzed groups of cells with significant correlation coefficients (P < 0.01)
In Situ Hybridization
Non-radioactive hybridizations were performed on four trkB–/– mice and five control littermates (P10–12), following the method described by Carmona et al. (2003). RNA probes against 65 kDa isoform of the GABA-synthesizing enzyme glutamic acid decarboxylase (GAD65) and KCC2 have been described previously (Aguado et al., 2003). Sections were hybridized overnight at 60°C with 50 ng/ml of GAD65 or KCC2 riboprobes labeled with digoxigenin-D-UTP (Roche Diagnostics, Germany). Thereafter, they were incubated with an alkaline phosphatase-labeled antidigoxigenin antibody (1:2000; Roche Diagnostics, Germany) and developed with a BCPI/NTB substrate. Sections from mutant and control mice were processed in parallel. Hybridizations with control sense riboprobes did not give any signal.
Electron Microscopy
Two control and two trkB–/– P12-13 mice were perfused with 1% glutaraldehyde and 1% paraformaldehyde in 0.12 M phosphate buffer. Brains were removed from the skull and fixed in the same solution overnight. Tissue slices were post-fixed with 2% osmium tetroxide, dehydrated and embedded in Araldite (Agar Scientific, Stansted, Essex, UK). Ultrathin sections were obtained and processed for post-embedding GABA immunostaining (Aguado et al., 2003) using a rabbit anti-GABA antiserum (1:2500; Sigma) and 15 nm colloidal gold-coated secondary antibodies (BBI International, Cardiff, UK). After staining with uranyl acetate and lead citrate, digital electron micrographs covering 67 µm2 (final magnification of 24 000x) were randomly captured from the stratum radiatum and the stratum pyramidale of the CA1 region (n = 45–52 micrographs per animal group). The density of GABA-positive contacts was calculated and compared in control and trkB mutant hippocampi. In control experiments, omission of the primary antibody or replacement with non-immunized IgGs prevented immunostaining.
Northern Blotting
Total RNA was obtained from the forebrain of two P10 trkB–/– and two wildtype (wt) littermates by the guanidine isothiocyanate method. Twenty micrograms of each sample was resolved in a denaturing 1.2% formaldehyde agarose gel and transferred to Hybond nitrocellulose membranes (Amersham). The membranes were prehybridized at 42°C for 2 h and then hybridized overnight at 42°C with GAD65 and KCC2 cDNA probes labeled with 32P (Aguado et al., 2003). The same filters were rehybridized with a mouse ß-actin cDNA probe to standarize RNA. Autoradiograms were captured with a scanner and densitometric analyses were performed using BioRad software.
Bicuculline-induced Hyperexcitability
Fura-2-loaded hippocampal acute slices from P9 trkB–/– mice and wt littermates (three and four animals, respectively), were continuously perfused in the recording chamber with oxygenated ACSF. For some slices, the standard ACSF was replaced by ACSF containing 30 µM of the GABAA inhibitor BMI, and network activity was recorded. Changes in [Ca2+]i were also registered from a set of slices after incubation with 0.5 µM BMI, and then again after further incubation with 30 µM BMI. Each slice was exposed to stimulation procedure once only.
Immunocytochemistry
trkB–/– mice and control littermates (P13; four wt and three mutant animals) were perfused with 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS), post-fixed in the same fixative and cryoprotected in 30% sucrose. Frozen coronal sections (40 µm thick) from mutant and wt mice were processed in parallel. After blocking endogenous peroxidases, free-floating sections were incubated in 5% serum–PBS for 1 h and then with anti-Fos antibody (1:5000; Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4°C. Sections were processed using the Vectastain ABC kit (Vector Laboratories, UK). Double staining immunocytochemistry was performed with primary antibodies anti-Fos (1:4000), anti-calmodulin kinase II (CaMKII) (1:1000; Chemicon, Temecula, CA), anti-calbindin and anti-parvalbumin (1:1000; Swant, Bellizona, Switzerland). After washes, sections were incubated with fluorochrome-conjugated antibodies (Alexa fluor 488 and Alexa fluor 568 anti-IgGs, 1:300; Molecular Probes). The brain sections were counterstained with bisbenzimide (Sigma) to analyze chromatin condensation. Some slices from Ca2+ imaging analysis were fixed in 4% paraformaldehyde at 4°C for 3 h and the interneuron identity of recorded cells was confirmed by the double labeling of fura-2 and calbindin. Incubation with nonimmunized IgGs and omission of primary antibodies prevented immunostaining. Fluorescent images were obtained with the Olympus Fluoview FV300 and Leica SPII scanning confocal microscopes.
Cell Counts
Brain sections processed for double immunolabeling were used for cell counting. The percentage of Fos-immunopositive neurons labeled for CaMKII, calbindin or parvalbumin was counted in wt and trkB–/– CA1 regions of equivalent hippocampal sections corresponding to the dorsal hippocampus. Two or three counts were performed for each animal (four wt and three trkB–/– animals).
Statistics
Student's t-test and the Mann–Whitney U-test were used for statistical evaluations. Data are expressed as mean ± SEM.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Spontaneous Correlated Network Activity is Decreased in Pyramidal Neurons of Neonatal trkB–/– Hippocampus
To investigate the in vivo participation of TrkB signaling in the postnatal development of spontaneous correlated activity, we analyzed [Ca2+]i changes in pyramidal neurons of the hippocampal CA1 region in trkB–/– brain slices (Ben-Ari et al., 1989; Garaschuk et al., 1998; Leinekugel et al., 2002; Aguado et al., 2003). Since activity levels and the mechanisms underlying spontaneous neuronal activity change markedly over the first postnatal days, we compared neonatal (P2–3) and juvenile (P8–9) mice. Spontaneous activity in hippocampal networks was determined in acute brain slices with the [Ca2+]i indicator fura-2 (Schwartz et al., 1998; Mao et al., 2001; Aguado et al., 2003).
After loading the dye, a vast number of cells were stained throughout the distinct hippocampal layers (Fig. 1A,B). Pyramidal neurons were identified under DIC optics on the basis of their cell body location within the stratum pyramidale, their distinctive triangular somata (15–18 µm in size) and their apical dendrites (Fig. 1A–D) (Freund and Buzsaki, 1996; Garaschuk et al., 1998). Optical recording of P2–P9 hippocampal slices over 800 s showed that 56% of pyramidal neurons (n = 10 slices from five animals) regularly displayed spontaneous changes in [Ca2+]i (wt in Figs 1E and 3A). These non-evoked Ca2+ events were entirely dependent on [Ca2+]o (97% blockage in nominally Ca2+-free, 2 mM EGTA, 0 mM [Ca2+]o ACSF; n = 67 cells from three slices). Moreover, incubation of hippocampal slices with 2 µM TTX virtually abolished spontaneous [Ca2+]i oscillations (91% blockade, n = 161 cells from 3 slices). These data indicate that spontaneous [Ca2+]i transients in postnatal pyramidal neurons occur regularly and depend on the activation of voltage-gated Na+ channels and Ca2+ influx, in agreement with earlier reports (Ben-Ari et al., 1989; Garaschuk et al., 1998).
|
|
P2–3 hippocampal pyramidal neurons from wt mice showed a robust spontaneous activity (Fig. 1E–G), as expected for the concomitant depolarizing actions of GABA and glutamate in neonatal mice (Ben-Ari et al., 1997
). Thus, most imaged pyramidal neurons of P2–3 hippocampi from wt mice (85.46 ± 6.46 imaged cells/movie) exhibited spontaneous [Ca2+]i oscillations (79 ± 5%) with a mean activation frequency of 43 ± 5 calcium transients/cell/104 s (seven slices from three animals; Fig. 1E–G) (Garaschuk et al., 1998; Aguado et al., 2003). A representative plot of F/F over time showing spontaneous [Ca2+]i transients in a P3 pyramidal neuron is illustrated in Figure 1E, in which the onset of each [Ca2+]i oscillation is labeled with a mark at the bottom of the graph. When spontaneous activity was analyzed in pyramidal neurons from P2–3 trkB-deficient littermates (79.2 ± 4.87 imaged cells/movie), no changes in the rate of activation were observed (42.76 ± 11.9 calcium transients/cell/104 s) (Fig. 1E,G). In contrast, a significant 32.5% decrease in the number of active pyramidal cells was detected in mutant slices compared with wt mice (53 ± 8% in trkB–/–; five slices from two animals; P = 0.01; Fig. 1F).
To analyze synchronous patterns of spontaneous activity generated by pyramidal assemblies, we used a previously described statistical method that identifies and maps coactivations among a large number of individual cells (Schwartz et al., 1998; Aguiló et al., 1999; Aguado et al., 2003). Highly synchronous patterns of spontaneous neuronal activity were routinely observed in the hippocampus of P2–3 wt mice, as previously described (Ben-Ari et al., 1989; Garaschuk et al., 1998; Aguado et al., 2003). A representative analysis of a P2 wt hippocampus is shown in Figure 2A–C. All fura-2-loaded pyramidal neurons that showed spontaneous [Ca2+]i changes are indicated with black squares. The onset of each [Ca2+]i oscillation (vertical marks) of each active neuron (horizontal lines) illustrated in Figure 2A is represented in a raster plot over time (66 cells) (Fig. 2B). Finally, the spatio-temporal correlation map (see Materials and Methods) shown in Figure 2C shows the synchronous network shaped by coactive neurons. In these maps, each pair of neurons with significant synchronous [Ca2+]i increases is connected by lines whose thickness is proportional to the degree of correlation. Consistently, almost every active neuron of each recorded wt hippocampus (93%, n = 7 slices from three animals) belonged to correlated networks (Fig. 2G). Moreover, coactivations recruiting >50% of active pyramidal neurons (arrows in raster plot) were frequently observed (22 cases in nine slices) (Fig. 2H).
|
Examination of the patterned spontaneous activity of P2–3 pyramidal neurons from trkB-deficient littermates revealed considerable alterations (Fig. 2D–F). Thus, the number of active pyramidal neurons belonging to correlated networks was 16% lower in mutant hippocampi (93.9 ± 2.43 in wt mice versus 79.03 ± 8.53 in trkB null; P = 0.035) (Fig. 2G). More importantly, although synchronous Ca2+ events occurred among pyramidal neurons (Fig. 2E,F), no synchronous coactivations recruiting >50% of active neurons were observed in trkB–/– hippocampi (n = 5 slices from two animals) (Fig. 2E,H). Therefore, we conclude that spontaneous correlated network activity is impaired in neonatal trkB–/– hippocampus.
Pyramidal Neurons of Juvenile trkB–/– Hippocampus Display Enhanced Spontaneous Correlated Network Activity
At the end of the first postnatal week, the time when GABAergic neurotransmission becomes inhibitory, levels of spontaneous network activity markedly decrease (Garaschuk et al., 1998; Ben-Ari, 2001, 2002). In agreement with these data, the percentage of spontaneously active pyramidal neurons in wt hippocampi fell from 79% in P2–3 (seven slices from three animals) to 33% in P8–9 (60.7 ± 1.98 imaged cells/movie from 18 slices from five animals) (Fig. 3B). When pyramidal neurons of P8–9 trkB–/– littermates were analyzed (63.26 ± 4.19 imaged cells/movie), a dramatic enhancement of spontaneous activity levels was detected. We found a 54% increase (P = 0.009) in the number of active pyramids in mutant mice compared with controls (51 ± 5% in trkB knockout mice; 19 slices from four animals) (Fig. 3B). Furthermore, the frequency of activation was 2.3-fold higher (P = 0.0019) in trkB–/– pyramidal neurons (42.02 ± 3.04 calcium transients/cell/104 s in wt versus 98.28 ± 13.97 in mutant pyramids) (Fig. 3A,C).
Next we analyzed network patterns of spontaneous activity in P8–9 trkB–/– hippocampi. In the wt hippocampus, synchronous network activity among pyramidal neurons was moderate (Fig. 4A–C). The proportion of pyramidal neurons belonging to correlated networks was 63% and the coactivations recruiting >50% of active cells (arrow in raster plot) were very low (19 cases in 18 slices) (Fig. 4G,H). In contrast, trkB-deficient hippocampi displayed high correlated activity (Fig. 4D–F). The proportion of recruited neurons into synchronous networks was significantly increased (74.8%, P = 0.032) and the number of coactivations recruiting more than half the active cells was notably enhanced (P = 0.014, 79 cases in 19 slices) (Fig. 4G,H).
|
Taken together, our data show that the lack of TrkB signaling during development causes a decrease in spontaneous excitability of pyramidal neuronal networks in neonatal animals but an increase in juveniles.
Increased Spontaneous Network Activity in Pyramidal Neurons of Juvenile trkB–/– Hippocampus is Associated with an Induction of Fos Transcription Factor
Neuronal excitability, mainly through changes in [Ca2+]i, controls the activation of specific transcription factors that initiate transcription and/or repression of other genes to achieve the long-term effects of neuronal activity (West et al., 2002). The protein product of the immediate early gene c-fos, Fos, is a well-known activity-dependent transcription factor involved in neuronal plasticity and pathology (Kash et al., 1997; Hughes et al., 1999; Woo et al., 2002). To further substantiate altered patterns of [Ca2+]i oscillation in trkB–/– mice, we examined Fos immunoreactivity in neurons of juvenile wt and mutant hippocampi. Because activity-dependent Fos expression is regulated developmentally and cannot be induced in vivo before P13 (Sakurai-Yamashita et al., 1991; Silveira et al., 2002), we used P13–14 wt and trkB knockout mice.
Consistent with previous observations (Sakurai-Yamashita et al., 1991; Silveira et al., 2002), baseline expression of Fos assessed by immunocytochemistry was absent in all layers of the CA1 region of the hippocampus of juvenile (P13) wt mice (Fig. 5A). In striking contrast, trkB–/– hippocampi showed a strong Fos induction through the CA1 pyramidal cell layer, with many cell nuclei robustly stained (Fig. 5B). To identify cell types expressing the Fos transcription factor in mutant hippocampus, we performed double labeling immunocytochemistry. Quantitative analysis of Fos and the specific pyramidal cell marker calmodulin-kinase II (CaMKII) showed that 22% of hippocampal pyramidal neurons from trkB knockout mice exhibited Fos immunostaining (Fig. 5C,D,H). As shown by Hoescht staining, the chromatin appearance of each Fos-positive nucleus was undistinguishable from negative nuclei (Fig. 5E). Furthermore, the lack of Fos immunoreactivity in calbindin- and parvalbumin-positive interneurons (Fig. 5F–H) indicated that Fos induction in trkB–/– hippocampus was specific for pyramidal neurons.
|
Because spontaneous Ca2+ activity in juvenile hippocampi were studied in P8–9 mice, we performed additional imaging experiments on P12 animals to ascertain the correlation between Ca2+ activity and Fos induction in mutant pyramidal neurons. After analysis of P12 mice (three slices from three animals from each group), we found that the percentage of spontaneously active pyramids was significantly higher (P < 0.01) in the mutant hippocampus compared with controls (trkB knockout mice: 40.16 ± 5.16% active pyramids, 15.3 ± 1.4 imaged cells/movie; control mice: 11.6 ± 1.3% active pyramids, 17.66 ± 1.85 imaged cells/movie). Moreover, a tendency towards an increase in the activation rate of trkB–/– pyramidal neurons (37.17 ± 7.5 calcium transients/cell/104 s in mutant pyramids versus 22.83 ± 5.45 in control neurons) was also observed.
Taking together, these results show a correlation between increased spontaneous Ca2+ activity and an aberrant induction of Fos transcription factor in pyramidal neurons of juvenile trkB–/– mice.
Impairment of Spontaneous Calcium Oscillations in Postnatal trkB–/– Hippocampal Interneurons
The results reported above show that elimination of TrkB signaling leads to a sequential decrease and an enhancement of spontaneous correlated activity in developing pyramidal networks. The striking correlation between the contrary consequences of trkB elimination on network activity and the opposite actions of GABA signaling during postnatal life (Ben-Ari, 2001, 2002) raises the possibility that the regulation of hippocampal network excitability by TrkB is mediated by altered function of GABAergic neurotransmission. To determine intrinsic activation levels of trkB–/– GABAergic neurons, we analyzed non-evoked [Ca2+]i oscillations in identified interneuronal somata of wt and trkB–/– hippocampal slices.
Interneuron cell bodies were identified under DIC optics by their typical large round-shaped somata (
20–25 µm in size) (Fig. 6A,B). The criterion for identifying interneurons was confirmed by the dual labeling of fura-2 and the expression of the specific marker calbindin (Freund and Buzsaki, 1996) (Fig. 6C,D). At neonatal stages, no significant changes were found in the percentage of spontaneously active internerneurons between the two groups of animals (88.1 ± 8.67% in wt mice, 23 imaged cells from seven slices from four animals; versus 80.55 ± 9.59% in mutant mice, 31 imaged cells from 12 slices from four animals) (Fig. 6G). However, activation rate of trkB–/– interneurons showed a 39.5% reduction compared with controls (77.28 ± 11.96 calcium transients/cell/104 s in wt versus 46.73 ± 6.15 in mutant interneurons; P = 0.01) (Fig. 6E,H). Similar results were observed in P8–9 hippocampal slices. At this age, the number of interneurons showing spontaneous [Ca2+]i oscillations was preserved in trkB–/– mice (44 ± 2% active interneurons in wt mice, 86 imaged cells from 24 slices from 13 animals; versus 38 ± 9% in mutant mice, 65 imaged cells from 13 slices from four animals) (Fig. 6G), but the frequency of activation was decreased by 43% (47.91 ± 6.31 calcium transients/cell/104 s in wt versus 27.21 ± 3.8 in mutant interneurons; P = 0.03) (Fig. 6F,H).
|
We conclude that trkB deletion preserves the number of spontaneously active interneurons, but consistently reduces their [Ca2+]i oscillation rates in both newborn and juvenile hippocampi.
Expression Levels of GAD65 mRNA and the Number of GABAergic Synapses are Markedly Reduced in Juvenile trkB–/– Hippocampus
Altered contents and synapses of the GABA neurotransmitter are associated with considerable changes in network excitability of both juvenile and adult cortical circuits (Kash et al., 1997; McCormick and Contreras, 2001). Therefore, we analyzed the expression of GAD65 and the number of GABA-positive synaptic contacts in the CA1 region of juvenile trkB–/– hippocampus.
Non-radioactive in situ hybridization assays for GAD65 mRNA showed that expression levels of the GABA-synthesizing enzyme were much lower in hippocampal interneurons of P10-11 trkB–/– mice than in wt littermates (Fig. 7A,B). A robust hybridization signal for GAD65 transcripts was observed in the somata of interneurons located in the CA1 region of the wt hippocampus (Fig. 7A). In contrast, hippocampal interneurons of trkB–/– mice showed a very faint signal for GAD65 mRNA (Fig. 7B).
|
To quantify the GABAergic inputs which contact CA1 pyramidal neurons, we performed an immunogold electron microscopy study of juvenile hippocampi. GABA-positive terminals containing gold particles and establishing synaptic contacts were counted in the stratum radiatum and the stratum pyramidale of wt and mutant hippocampi. We noted a 42% reduction of GABAergic terminals contacting postsynaptic dendrites in the stratum radiatum of mutant mice (P < 0.01) (Fig. 7C,D,F). Interestingly, the greatest reduction was detected in the number of GABA-positive terminals inervating the perisomatic region of pyramidal cells in the stratum pyramidale in trkB–/– mice (55% decrease with respect to wt animals, P < 0.01) (Fig. 7E,F).
We conclude that hippocampal microcircuits of postnatal trkB–/– mice display a dramatic decrease in both the expression of GAD65 and the number of GABAergic synaptic contacts onto pyramidal neurons, indicating an impairment of GABAergic neurotransmission in trkB mutant mice.
Inhibitory Action of GABAergic Neurotransmission is Preserved in Juvenile trkB–/– Hippocampus
In the hippocampus, GABAergic neurotransmission through GABAA receptors switches from depolarizing to hyperpolarizing at the end of the first postnatal week (Ben-Ari et al., 1989). Thus, the enhancement of network activity observed in juvenile trkB–/– hippocampi could be the consequence of an ineffective conversion of GABAergic function from excitatory to inhibitory. Thus, we analyzed the action of GABAA receptor-mediated neurotransmission in juvenile mutant mice.
It has been shown that the GABAergic switch depends on the developmentally-regulated expression of the K+/Cl– co-transporter KCC2 (Rivera et al., 1999; Hübner et al., 2001). To determine the effects of GABAA receptors on synaptic function in juvenile trkB–/– mice, we first evaluated the expression of the KCC2 cotransporter in P10-12 mutant animals (n = 2 each group). Northern blot analysis of total forebrain RNA revealed a reduction of the KCC2 mRNA expression in trkB mutant mice, compared with control littermates (Fig. 8A). Similarly, a moderate decrease in the KCC2 hybridization signal was detected in the neocortex and hippocampus of trkB–/– mice by non-radioactive in situ hybridization (Fig. 8B,C).
|
Next, we determined the action of GABAergic neurotransmission in juvenile trkB mutant mice by analyzing changes in spontaneous network activity after bath application of the GABAA receptor blocker BMI. In agreement with previous studies (Leinekugel et al., 1997
; Garaschuk et al., 1998; Aguado et al., 2003), incubation of hippocampal slices with 30 µM BMI markedly decreased spontaneous [Ca2+]i oscillations in P2–3 wt mice (data not shown). These observations are consistent with the excitatory actions of GABA in the neonatal hippocampus (Ben-Ari, 2001). In contrast, pharmacological blockade of GABAA receptors by 30 µM BMI elicited epileptiform network activity in wt juvenile hippocampus. During treatment almost every fura-2-loaded neuron became active (94%; four slices from two animals) (Fig. 9A,B) and the spatio-temporal correlations among [Ca2+]i oscillations were markedly enhanced (see correlation maps in Fig. 9A). Moreover, the number of Ca2+ events recruiting >50% of the active pyramidal neurons was increased in each sample (Fig. 9C). As previously reported (Albowitz et al., 1997; Badea et al., 2001; Aguado et al., 2002), this enhancement of network activity elicited by BMI demonstrated the inhibitory action of GABA in juvenile wt hippocampus. In good agreement with the above results (Figs 3 and 4), the basal network activity of trkB mutant hippocampi was increased compared with slices from wt littermates (correlation maps in Fig. 9A and histograms in Fig. 9B,C). Notably, blockade of GABAA receptors with 30 µM BMI activated most pyramidal neurons recorded in trkB–/– hippocampal slices (89.54 ± 4.74%; 5 slices from 2 animals) (Fig. 9A,B). Furthermore, the occurrence of coactivations recruiting >50% of active neurons in BMI-treated knockout hippocampi was similar to BMI-treated wt slices (6 ± 1.35 coactivations/movie in trkB–/– and 5.4 ± 1.12 coactivations/movie in wt) (Fig. 9C). This similar effect on network activity registered after pharmacological blockade of GABAA receptors in trkB–/– and wt slices indicates that diminished levels of KCC2 mRNA in the juvenile mutant hippocampus were high enough to sustain the inhibitory conversion.
|
The above data showing that conversion of GABAergic neurotransmission is preserved in postnatal trkB–/– hippocampus suggest that the increased spontaneous network activity detected in juvenile mutant pyramidal neurons is due to an impairment of inhibitory GABAergic transmission, rather than to a failure in its developmental switch.
Postnatal trkB-deficient Hippocampus shows Enhanced Susceptibility to Network Hyperexcitability
In the juvenile and adult hippocampus, the fine-tuned adjustment of inhibitory GABAergic neurotransmission is essential to drive distinctive activity patterns of pyramidal neurons, and prevents network hyperexcitability (Freund and Buzsaki, 1996; Kash et al., 1997; McBain and Fisahn, 2001; Klausberger et al., 2003). The consistent impairment of spontaneous [Ca2+]i oscillation in trkB–/– interneurons, together with the changes in components of the GABAergic system (GABA-positive synaptic contacts, and GAD65 and KCC2 expression), suggests that increased network excitability of juvenile mutant hippocampus is caused by a reduction in the inhibitory strength of GABA neurotransmission.
To analyze the strength of endogenous inhibitory GABAergic neurotransmission in juvenile trkB–/– mice, we incubated juvenile hippocampal slices with very low concentrations of the GABAA receptor antagonist BMI. First, we assayed the highest concentration of BMI at which patterns of spontaneous activity in wt hippocampus were unaltered. In agreement with previous studies (Smart et al., 1998; Gabel and LoTurco, 2002; Menendez de la Prida and Pozo, 2002), low dosages of BMI, which minimally reduce but do not completely block GABAA-mediated inhibition, had virtually no effect on the pattern of spontaneous [Ca2+]i oscillation of wt pyramidal neurons (Fig. 10A,B). As shown in representative raster plots and histograms (Fig. 10A,B), no changes were observed in the number of active pyramidal neurons or in their activation frequency in 0.5 µM BMI-treated wt hippocampus (n = 7 slices from three animals). In addition, the correlation degree among active pyramidal neurons was also preserved in wt hippocampi incubated with a low concentration of BMI (Fig. 10A,B). Because replacement of 0.5 µM BMI by 30 µM BMI caused network hyperexcitability (see Fig. 9) in every treated wt slice (data not shown), we ruled out the incapacity of these wt hippocampi to generate epileptiform network activity.
|
In contrast, the partial antagonism of GABAA inhibition produced by 0.5 µM BMI in trkB–/– hippocampal circuits resulted in a marked increase in the excitability of pyramidal populations (n = 5 slices from three animals) (Fig. 10A,B). We observed significant 2.5- and 1.7-fold increases in the number of active pyramidal neurons and in the frequency of [Ca2+]i oscillations, respectively, compared with the basal condition (Fig. 10B). Furthermore, network synchrony of mutant hippocampi was also enhanced, with a 2-fold increase in the coactivation events involving more than half the active pyramidal neurons (arrows) (Fig. 10A,B).
These data show that the juvenile trkB–/– hippocampus has a lower threshold for BMI-induced network hyperexcitability. Taken together, our results suggest that the lack of TrkB signaling during development impairs the strength of GABAergic synaptic inhibition, thereby leading to increased excitability of neuronal circuits.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
GABAergic Synaptogenesis and Developmental Expression of KCC2 In Vivo Depend on the trkB Gene
The present study shows that genetic inactivation of the tyrosine kinase domain of trkB throughout development impairs pre- and post-synaptic components of the GABAergic neurotransmission. At the presynaptic level, ultrastructural and transcriptional analysis of postnatal trkB–/– mice revealed notable deficiencies in the expression of the GABA synthesizing enzyme GAD65 and the density of GABA-positive synaptic contacts. Given that BDNF increases GAD expression and the density of GABAergic synapses in the developing CNS (Bolton et al., 2000; Marty et al., 2000; Seil and Drake-Baumann, 2000; Richardson and Leitch, 2002; Yamada et al., 2002; Aguado et al., 2003), the present results indicate that BDNF controls GABAergic synaptogenesis in vivo via TrkB signaling.
Changes in the postsynaptic components were also detected in the GABAergic system of trkB-deficient hippocampi. Thus, transcriptional expression of the K+/Cl– cotransporter KCC2 was decreased in the mutant hippocampus. These results are consistent with the opposite effect observed in BDNF-overexpressing transgenic embryos and with the notion that GABAergic neurotransmission adjusts KCC2 expression (Ganguly et al., 2001; Aguado et al., 2003). In contrast to genetic approaches, a down-regulation of KCC2 expression and activity has been reported after brief administration of BDNF in hippocampal cultures and brain slices (Rivera et al., 2002). These differences could be related to the dependence of BDNF effects on temporal and developmental aspects, as previously proposed (Lu, 2004). Moreover, although indirect effects of TrkB signaling on KCC2 expression cannot be ruled out, the consistent results obtained from genetic models indicate that the developmental rise of KCC2 in vivo depends on TrkB signaling.
TrkB Signaling Regulates Interneuron Activity and Network Excitability in Postnatal Mice
Although it has been shown that trkB regulates development of GABAergic components (Bao et al., 1999; Huang et al., 1999; Marty et al., 2000; Rico et al., 2002; Aguado et al., 2003, Carmona et al., 2003), the functional consequences of TrkB signaling on interneuron physiology and GABAergic neurotransmission in vivo is poorly understood. Here we show that deletion of the trkB gene dramatically reduced the rate of spontaneous [Ca2+]i oscillations in hippocampal interneurons, in both newborn and juvenile mice. This in vivo observation is consistent with previous electrophysiological studies that show that long-term administration of BDNF enhances spontaneous activity in cultured cortical interneurons (Rutherford et al., 1998; Bolton et al., 2000). Since the composition of glutamate and GABAA receptor subunits in interneurons is TrkB-dependent (Narisawa-Saito et al., 1999; Carmona et al., 2003), it is feasible that an aberrant function of neurotransmitter receptors underlies activity alterations of mutant interneurons. Thus, TrkB signaling in vivo is required not only to adjust the number of GABAergic synapses, but also to regulate interneuron activation patterns.
Strikingly, the GABAergic synaptic alterations observed in postnatal trkB–/– mice were correlated with an increased susceptibility to network hyperexcitability. Thus, the limited blockade of GABAA receptors by low dosages of BMI distinctively triggered network hyperexcitability in trkB–/– hippocampi. This increased susceptibility to hyperexcitability in juvenile trkB–/– mice suggests that the inhibitory strength of GABA is reduced in the absence of TrkB signaling in vivo. Since the 65 kDa isoform of GAD is the major enzyme synthesizing GABA for neurotransmission and perisomatic GABAergic synapses governs the output of pyramidal neurons (Kash et al., 1997; Klausberger et al., 2003), it is likely that the impaired pre-synaptic GABAergic components in mutant mice are not enough to prevent hyperexcitability under partial disinhibition. Moreover, although the inhibitory action of GABA was essentially preserved in juvenile mutant hippocampi, the reduced expression of KCC2 in trkB–/– mice may contribute to enhancing neuronal excitability by impairing Cl– extrusion (Rivera et al., 1999; Hübner et al., 2001; Woo et al., 2002).
TrkB Signaling Adjusts the Intrinsic Activity of Developing Networks by Regulating GABAergic Neurotransmission
We observed that spontaneous network activity in the wt hippocampus was very robust early after birth, but notably reduced in juvenile mice. Although network correlation among active neurons was maintained, synchronous [Ca2+]i transients, connecting the entire neuron population, were lost during the second postnatal week. These data agree with previous optical and electrophysiological studies performed in the postnatal hippocampus (Ben-Ari et al., 1989; Garaschuk et al., 1998; Ben-Ari, 2001; Aguado et al., 2003). Interestingly, spontaneous correlated network activity of pyramidal neurons was impaired in neonatal trkB–/– mice, but significantly enhanced at juvenile stages. The decreased activity in neonatal trkB–/– hippocampus is consistent with a previous report showing that BDNF-overexpressing embryos display increased activity (Aguado et al., 2003) and points to a crucial role in vivo for BDNF/TrkB in the emergence of the earliest cortical activity.
Our observations of an enhancement of spontaneous [Ca2+]i oscillations in hippocampal networks of juvenile trkB–/– mice may conflict with the proposed contribution of BDNF to adult epileptogenesis (Binder et al., 2001). However, although the role of TrkB signaling in increasing network excitability is established for acute BDNF treatments, the role of chronic TrkB activation on epileptic susceptibility is a matter of controversy (Reibel et al., 2001). Strikingly, the correlation between Fos expression and the levels of Ca2+ activity detected in juvenile trkB–/– hippocampi substantiates the hyperactive status of mutant neuronal networks. It is noteworthy that similar Fos responses have been reported in the postnatal hippocampus after pharmacological induction of seizures (Sakurai-Yamashita et al., 1991; Silveira et al., 2002; Zhang et al., 2002). Furthermore, our results are in agreement with the fact that evoked excitability is decreased in the visual cortex of postnatal mice overexpressing BDNF (Huang et al., 1999). Regarding the restricted induction of Fos detected in trkB–/– pyramidal neurons (22%), specific Ca2+ activity patterns may activate Fos expression in subsets of spontaneously active cells (Dolmetsch et al., 1998; Li et al., 1998).
The rapid depolarization and potentiation of [Ca2+]i oscillations, previously shown after acute BDNF administration, may indicate direct regulation of network activity by endogenous TrkB signaling (Kafitz et al., 1999; Numakawa et al., 2002). However, the opposing effects on intrinsic network activity observed in neonatal and juvenile trkB–/– pyramidal neurons suggest a more complex regulation. Pyramidal neurons development was the first BDNF target identified in the CNS (see Huang and Reichardt, 2001; McAllister, 2001). In agreement with this, impaired axonal and synaptic development has been reported in pyramidal neurons of the postnatal trkB–/– hippocampus (Martínez et al., 1998; Carmona et al., 2003). Although an altered glutamatergic system may contribute to changes in network physiology in the trkB–/– hippocampus, the age-dependent effects of trkB elimination on patterned activity make it unlikely that glutamatergic transmission is the main target of trkB deletion. In contrast, the present functional and structural deficiencies detected in the GABAergic system of trkB–/– mice, together with the opposite actions of GABA d