| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Cell and Neurobiology, Humboldt University Hospital (Charité), Institute of Anatomy, 10115 Berlin, Germany;
* Department of Neurobiology, German Primate Center, 37077 Göttingen, Germany; and
Department of Psychiatry, UT Southwestern Medical Center, Dallas, Texas 75390, USA
2Correspondence: Institute of Anatomy, Department of Cell and Neurobiology, Humboldt University Hospital (Charité), Philippstr. 12, 10098 Berlin, Germany. E-mail: robert.nitsch{at}charite.de
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
21 days after lesion (dal). Immunohistochemistry demonstrated a corresponding regulation of HCN1 protein expression in CA1-CA3 dendrites, hilar mossy cells and interneurons, and granule cells. Patch-clamp recordings in the early phase after lesion from mossy cells and hilar interneurons revealed an increase in the fast time constant of current activation and a profound negative shift in voltage activation of Ih. Whereas current activation recovered at 30 dal, the voltage activation remained hyperpolarized in mossy cells and hilar interneurons. Granule cells, however, were devoid of any detectable somatic Ih currents. Hence, denervation of the hippocampus decreases HCN1 and concomitantly the Ih activity in hilar neurons, and the recovery of h-current activation kinetics occurs parallel to postlesion sprouting.—Bräuer, A. U., Savaskan, N. E., Kole, M. H. P., Plaschke, M., Monteggia, L. M., Nestler, E. J., Simbürger, E., Deisz, R. A., Ninnemann, O., Nitsch, R. Molecular and functional analysis of hyperpolarization-activated pacemaker channels in the hippocampus after entorhinal cortex lesion.
Key Words: DDRT-PCR/ECL • HCN • hippocampus • Ih
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3
4
5)
. The EC provides a dual input to the hippocampus: axons from layer II project to the dentate gyrus (DG) and axons from layer III project to the stratum-lacunosum moleculare of the CA1-CA3. The EC projection to the DG terminates along the outer two-thirds of the molecular layer of the DG whereas, sharply segregated, the inner one-third is innervated by excitatory commissural and associational fibers. The input arriving at the distal two-thirds of the granule cell (GC) dendrites is conveyed by GC axons (mossy fibers) to the dendrites of CA3 pyramidal neurons (6
, 7)
, which in turn excite the pyramidal cells in CA1 via Schaffer collaterals. Mossy cells (MC) and hilar interneurons (HI) act as modulators for this trisynaptic information flow (8)
. Whereas MCs are excitatory and give rise to fibers of the associational/commissural pathway, HIs are predominantly inhibitory and project back to the granule cell dendrites and their somata in a laminated fashion, which indicates a pivotal role to modify inputs from the entorhinal region (9
, 10)
. Hilar neurons are thought to pace the subthreshold oscillations in the hippocampus (11
, 12)
, hence modulating the information flow between the isocortex and the hippocampus proper.
Transection of the perforant path via entorhinal cortex lesion (ECL) results in an anterograde degeneration of entorhinal fibers and synapses in the outer molecular layer of the DG and in the stratum-lacunosum moleculare of CA1 and CA3 (2
, 3
, 13)
. This process is followed by reactive sprouting leading to the reappearance of intact synapses of almost normal levels, indicating the replacement of the entorhinal synaptic input by other fiber systems from the septum and commissural/associational projections (14
15
16)
. The sprouting response occurs in a layer-specific manner also involving unlesioned GABAergic fibers replacing the lesioned glutamatergic entorhinal fibers (14
, 15
, 17
, 18)
, resulting in an imbalance of excitation and inhibition in the DG after lesion (19)
.
In our attempt to identify the molecular factors involved in structural and functional changes of the hippocampus after deafferentation, we performed a differential display screen for genes whose transcription is altered after ECL. We identified a gene encoding a hyperpolarization-activated pacemaker channel, HCN1, which was down-regulated immediately after ECL. We focused further investigations on neurons of the DG, the first relay station in the hippocampal trisynaptic pathway. The ECL-induced down-regulation caused a transient decrease in detectable HCN1 protein and HCN1-like current (Ih) in MCs and interneurons of the hilar region. These transneuronal changes of the Ih properties in the hilus might contribute to functional changes in the hippocampus after ECL, suggesting an activity-dependent regulation of the HCN1-mediated Ih.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
.
Fluoro-Jade and acetylcholinesterase staining
To confirm the completeness of the ECL, we used Fluoro-Jade staining as a neurodegeneration marker for early survival stages from 1 to 15 dal. The Fluoro-Jade (Histo-Chem, Jefferson, AR) staining is adopted from Schmued et al. (21)
with some modifications (22)
. Acetylcholinesterase histochemistry (AChE) was performed with ECL stages from 20 and 30 dal, as described (23)
.
Differential mRNA display and cloning
The hippocampi from 1 dal and control animals (n=6) were rapidly removed and washed in 0.1 M phosphate buffer (PB). Total RNA was extracted using the guanidinium thiocyanate-phenol-chloroform extraction method (24)
. DNA-free RNA was obtained by treatment with RNase-free DNase I (Boehringer, Mannheim, Germany) for 30 min at 37°C. After phenol-chloroform extraction and ethanol precipitation, the probes were stored at -80°C. In 20 µl 1x transcription buffer, the first strand reactions contained 50 ng Oligo (dt11-VN), 400 units of SuperScript II RT-transcriptase, 100 nM DTT (all from Life Technologies, Eggenstein, Germany) and were incubated with 2 µg total RNA and 200 µM dNTPs (Pharmacia, Erlangen, Germany) for 1 h at 42°C. After first strand synthesis, two units of RNase H were added and the reactions were incubated at 37°C for 30 min. The cDNA were purified using the PCR Purification Kit (Qiagen, Chatsworth, CA).
PCR was performed using 2 µl of the RT reaction per 25 µl. All reactions were performed in quadruple. The primer used anchor and arbitrary primer (Roth) in different combinations. Reactions were performed in 25 mM MgCl2, 2.5 µl 10x reaction buffer, 2.5 units AmpliTaq Gold (all from Perkin Elmer, Norwalk, CT), 200 mM dNTPs (Pharmacia Biotech), 20 pmol per anchor primer, and 10 pmol per arbitrary primer. The thermal cycler was set for 1 cycle of 10 min at 95°C, 90 s at 40°C, and 45 s at 72°C, followed by 39 cycles of 30 s at 94°C, 90 s at 40°C, and 45 s at 72°C. The cycling was followed by 5 min at 72°C, then soaked at 4°C. For electrophoretic separation, the whole PCR reaction was ethanol precipitated. For analysis of PCR products, the CleanGel DNA analysis kit (Pharmacia Biotech) and the DNA silver staining kit (Pharmacia Biotech) were used according to the manufacturer’s instructions. Reamplification of the bands differing in their expression between lesion and control animal were excised and reamplified in 25 µl reaction using the same conditions as before. The resulting fragments were cloned into the pCR®2.1-TOPO Vector (Invitrogen, San Diego, CA) and sequenced.
Kainic acid-induced seizures
Male Wistar rats received a single s.c. injection of kainic acid (12 mg/kg, Sigma, Taufkirchen, Germany) as described earlier (25)
. The effects of seizure activity were measured by the criteria of Bendotti et al. (26)
. After 6 h or 24 h, brains from the kainic and control (saline injected) animals were dissected and prepared for in situ hybridization as described above.
In situ hybridization
Six brains per lesion stage were dissected and frozen on dry ice. Horizontal cryostat sections (19 µm) were fixed in 4% paraformaldehyde (w/v), washed in 0.1 M PB (pH 7.4), and dehydrated. An antisense oligonucleotide (5' -CCG ATC GAG TCG GTC AAT AGC AAC TGT CTC AAA GGC TCT TCT CAT CAT-3') complementary to bases 1923–1970 of HCN1 cDNA (GenBank acc. # AJ225123) was used and gave results similar to those of the riboprobe for HCN1. The specificity of the oligonucleotide was confirmed by a BLAST GenBank search and showed no significant cross matches with any known sequences and ESTs. The in situ hybridization was performed as described in Bräuer et al. (20)
. The oligonucleotides were end labeled using terminal deoxynucleotide transferase (Boehringer Mannheim, Germany) and [
-35S]dATP (DuPont NEN, Boston, MA); 400,000 cpm labeled oligonucleotide in 50 µl hybridization buffer was used per section. Hybridization was performed for 16 h at 42°C in a humidified chamber, then the slides were washed: 2 x 30 min in 0.1x SSC at 56°C and 1 x 10 min in 0.05 x SSC at room temperature. Finally, the sections were rinsed in H2O at room temperature and dehydrated. For autoradiography, slides were exposed to Kodak X-OMAT AR X-ray films for 5 days. No signals were detected on sections hybridized with a specific antisense oligonucleotide when unlabeled oligonucleotide was added in 100-fold surplus. After this, the slides were dipped in Kodak NTB-2 emulsion and exposed for 3 wk at 4°C, then developed in Kodak D19 developer and counterstained with hemalaun-eosin. The riboprobes to the four HCN rat genes were generated to the nonconserved amino terminus of each channel (27)
. Each fragment is 250 bp with a similar GC content. In situ hybridization of the riboprobes was performed as described by Monteggia et al. (27)
.
Quantification of autoradiography
Six animals were analyzed by in situ hybridization at each of the four postlesion time points and the unlesioned control values. The quantification was performed as described by Bräuer et al. (20)
. For analysis of these images, a computerized videodensitometry system (Metamorph, Universal Imaging, Downingtown, PA) was used to quantify the signal intensity on a pixel level. A visually established pixel intensity threshold was set to remove the unlabeled portion of the image. A standard rectangle (1.5 mm2) was defined and placed at 10 different positions over the pyramidal cell layers of the CA1 region, hilus, and granule cell layer of the DG. A mean value of the 10 measurements was calculated for each region and set in reference to the unlesioned controls. Analysis was performed using the Mann-Whitney U test. The level of significance was set at P < 0.0001.
Immunostaining
The antisera aq1 and aq2 were generated by Santoro et al. (28)
against two separate domains in the predicted cytoplasmatic tail of mHCN1: amino acids 594–720 (antiserum aq1) and 777–910 (antiserum aq2). For immunocytochemical staining, three animals were used per lesion stage. Immunohistochemistry was performed as described by Santoro et al. (28)
: 20 µm cryostat sections of rat brain (fixed in 4% paraformaldehyde/PBS) were quenched in 50 mM NH4Cl/PBS for 30 min. The sections were blocked with 10% goat serum/0.1% Saponin in PBS for 1 h at room temperature and exposed to aq1 and aq2 antisera (diluted 1:400 in blocking solution) overnight at 4°C. After washing in PBS/0.1% saponin, sections were incubated with anti-rabbit biotinylated antisera for 4 h at room temperature and then in avidin-biotin peroxidase complex reagent (Vectastain ABC Kit; Vector Laboratories, Burlingame, CA) diluted 1:100 in 0.1 M PB, for 1.5 h at room temperature. The immunoreaction was visualized with 3,3'-diaminobenzidine as a chromogen (0.07% DAB and 0.002% hydrogen peroxide in 0.1 M PB) incubated for 10 min.
Electrophysiological recording
Animals were deeply anesthetized with ether and transverse hippocampal brain slices (400 µm) were prepared by standard methods. Artificial cerebrospinal fluid (ACSF) contained (in mM) 124 NaCl, 2.5 KCl, 1.25 NaH2PO4, 2 MgSO4, 26 NaHCO3, 10 D-glucose, and 2 CaCl2 (Merck, Darmstadt, Germany) and was bubbled with 95% O2 and 5% CO2. A modified ACSF was used for voltage clamp in which 20 mM of NaCl was replaced with 20 mM of TEA-Cl, NaH2PO4 was omitted, and 0.5 µM of TTX and 0.3 mM of CdCl2 (Sigma) were added. CNQX (30 µM; Tocris, St. Louis, MO) was added to inhibit excitatory synaptic activity. In some experiments where Ba2+ was used, 2 mM of MgSO4 was replaced with 2 mM of MgCl2. ZD7288 (Tocris) was freshly dissolved in ACSF to 10 mM and added to the external solution in the appropriate concentration. Patch-clamp recordings (29)
were performed on granule or hilar cells in the deep hilus in slices that were kept submerged at 32°C in a recording chamber equipped for IR-DIC videomicroscopy (30)
. Borosilicate glass capillaries (1.5 mm o.d.; Hilgenberg, Malsfeld, Germany) were pulled to final resistances between 3 and 6 M
(as measured in the bath) and filled with an intracellular solution composed of (in mM) 120 K-gluconate, 20 NaCl, 1 MgCl2, 1 CaCl2, 11 EGTA, 10 HEPES, and 2 Mg2+-ATP (Sigma). Whole-cell recordings in current-clamp mode (to assess cell quality by action potential heights and membrane potential) and voltage-clamp mode were done using a EPC-9 amplifier and PULSE software v8.11 (HEKA, Lambrecht, Germany). Series resistance was between 12–18 M, compensated to a maximum of 65%.
Cell identification
The large variety of neurons belonging to the hilar region may be physiologically divided into spiny (with the MC as the prototype) and ‘aspiny’ neurons, which include most of the HIs (31
, 32)
. A set of physiological and morphological criteria was used to identify the cell type in the hilus. Preliminary identification of MCs and HIs by their distinct features of somatic shape and size in the deep hilus was aided by the use of infrared videomicroscopy (10)
. Basket cells, which lack Ih (10)
, could be excluded from the recording because of their adjacent position at the granule layer and large somata. Moreover, basket cells were readily recognized electrophysiologically by the combination of a low input resistance and fast spiking pattern. The spiny MCs and aspiny HIs were further identified physiologically by their spike patterns. Previous studies in the hippocampus of adult rats using post hoc neuromorphological cell identification have consistently shown that MCs exhibit fast afterhyperpolarization (fAHP) with relatively small amplitudes (<10 mV) after a single action potential or a burst of action potentials (10
, 32
, 33)
. In contrast, spiny and aspiny HIs both possess large (>10 mV) fAHPs or slow AHP. To assess the fAHP for our experiments, we measured the voltage amplitude directly after action potential relative to action potential threshold (see Fig. 5A
). Cells were divided into two groups. The separation procedure yielded MC (n=22) with fAHP of 6.6 mV (range 5.7–7.5, n=4) and postlesion 6.7 ± 1.6 mV (range 2.0–10.0, n=18). The HI (n=32) had fAHP amplitudes of 12.2 ± 1.3 mV (range 10.4–14.6, n=7) and postlesion 13.8 ± 0.8 mV (range 10.3–17.1, n=25) that differed significantly (P < 0.02 and P < 0.01 at control and postlesion, respectively). Additional independent distinction between spiny MCs and aspiny interneurons derives from their differences in decay kinetics of spontaneous excitatory postsynaptic activity (32)
. Voltage recording of hilar cells at -50 mV and -60 mV (see Fig. 5A
) regularly contained mEPSCs, allowing quantification of the decay kinetics by fitting the decay rate of eight synaptic events per cell with single exponential functions. In HIs, the decay rate was on average 4.5 ± 0.5 ms (range 1.8–7.7, n=22) vs. 12.1 ± 2.2 for MCs (range 5.4–33.7, n=12), and groups were different during pre- and postlesion stages (P < 0.04 and P < 0.02, respectively). The combination of somatic appearance, mEPSCs kinetics, action potential, and AHP thus allowed for sufficient characterization of spiny MCs and HIs.
|
Resting input resistance (RN) was obtained from the slope of a linear fit of the plotted I-V curve (within+10 and -10 mV) of current-clamp recordings. Membrane time constant (
m) was estimated as the slowest value of single exponential fits of the rising phase of small depolarizations without synaptic activity (400 ms). The voltage ‘sag’ was calculated as the ratio of the inward peak voltage and end voltage elicited by a current injection of -200 pA. Voltage activation of Ih was determined using the slowly decaying tail currents from preceding hyperpolarizing steps and data were analyzed by a Boltzman fit:
 |
25 ms, and normalized to the maximum tail current (A2) from a -140 mV step. The data were plotted against the preceding step potentials (V), where V1/2, the midpoint of voltage activation and k, the slope of the curve, were obtained. Hyperpolarizing command steps of 500 ms and 1 s were used. Comparison of the two durations revealed that at 500 ms, 94% of the amplitude is attained at 1 s. Since some of the neurones deteriorated with the longer pulses, we confined routine measurements to 500 ms. A difference of 6% from the steady state was tolerated. Ih activation rate was determined by single and double exponential fits to the current rise according to the formula:
 |
1 is the current rate of current activation. The amplitude A0, the offset, was obtained from a back-extrapolation to the start of the current segment. A1 was used as the amplitude of the h-current (see also ref 34
). A2 and 2 were used when biexponentials were applied. Though most rising phases of the h-currents at -140 mV steps could be fit well using single exponential fits, biexpontential current activation, which has been reported for Ih in interneurons (35
, 36)
, only slightly improved the fit in a subset of neurons under the experimental conditions used in this study. Allowing quantitative group comparison, both measurements have been included in the analysis. The linear leakage and uncompensated capacitance were routinely recorded with a P/5 protocol. Analysis of Ih in hilar and dentate neurons was obtained from unsubtracted traces. To determine the instantaneous current, the offset current A0 was subtracted by the leak current amplitude. All reported fittings were deemed successful by visual inspection. Origin v 6.1 (Microcal Software, Northhampton, MA) was used for the graphical processing and sigmoidal fitting. Exponential fits were performed with PULSEFIT (HEKA). ANOVA, Tukey HSD post hoc tests, two-tailed t tests, and Pearson bivariate correlation tests were performed with SPSS v7.5 (SPSS Inc., Chicago, IL). Data are given as mean ± SE.
Electron microscopy
For ultrastructural studies, HCN1 immunoreactive sections were osmicated (1% OsO4 in 6.84% sucrose/PB for 5 min), dehydrated in graded ethanol, and flat-embedded in Epon between silane-coated slides. Ultrathin sections were cut on a Reichert Ultratome, mounted on single slot grids coated with Formvar film, and stained with lead citrate and uranyl acetate. A Zeiss EM 900 electron microscope was used for examination.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
A. Fibers from the EC and those from the contralateral hippocampus terminate in a topographically organized fashion along the dendrites of dentate granule cells (6)
. The segregated termination of entorhinal fibers is found in the outer two-thirds of the molecular layer of the DG (Fig. 1A
). To ascertain the completeness of ECL in animals with short survival times (i.e., 2 days after lesion, or 2 dal), we performed staining of sections with Fluoro-Jade, a dye that specifically stains degenerating neurons (22)
. As illustrated in Fig. 1B
, degenerated entorhinal fibers are marked with Fluoro-Jade located in the inner molecular layer of the DG and in the stratum-lacunosum moleculare of CA1-CA3 region. For longer survival stages (i.e., 30 dal), brain sections were stained for acetylcholine esterase (AChE). In control animals, AChE-positive fibers are ubiquitously present in all molecular layers of the DG (Fig. 1C
). Thirty days after ECL, AChE-positive staining is increased in the outer molecular layer of DG, indicating a profound reactive sprouting reaction of AChE-positive fibers into the deafferented region (Fig. 1D
).
|
Screening for genes differentially regulated after deafferentation
To identify genes whose level of expression is altered by ECL, we used the differential display reverse transcription polymerase chain reaction (DDRT-PCR) method (37)
. One PCR product of 92 bp, which was absent in lanes from RNA from deafferented hippocampi, could be amplified from the RNA of unlesioned control rats (Fig. 2
). The sequence of this clone (B07) revealed 98% homology to a mouse gene (GenBank acc. # AJ225123; 1902 bp -1989 bp) and 100% to a rat gene (GenBank acc. # AF247450; 1748 bp -1839 bp), the hyperpolarization-activated, cyclic nucleotide-sensitive cation nonselective channel (HCN1) (28
, 38
, 39)
.
|
HCN in situ hybridization
To evaluate the distribution of altered expression in the rat hippocampus, in situ hybridization was performed. In untreated control rats, in situ hybridization with the HCN1 riboprobe revealed strong hybridization signals in the CA1-CA3 pyramidal cell layers, the granule cell layer of the DG, and, to a lesser intensity, the hilar region (Fig. 3
A, A.1, A.2). One day after lesion, a decreased hybridization signal could be observed in all hippocampal layers (Fig. 3
B, B.1, B.2). The intensities were decreased by 80% in the hilus, 70% in the DG, and 45% in the CA1 region (Fig. 3D
). A slow recovery of HCN1 mRNA levels was detectable at 15 dal. At 21 dal (Fig. 3C
), normal expression levels were present in all hippocampal areas when compared with the controls. Brain regions of the contralateral hippocampus displayed no significant alteration in HCN1 expression levels compared with the controls (data not shown). Based on these data, we studied possible changes of other known members of the HCN family in the hippocampus. As previously reported, the expression pattern of HCN2/3/4 showed a distinct pattern of distribution (Fig. 3E
, G, I; 27
, 40
). Surprisingly, the mRNAs for HCN2/3/4 showed no significant changes after ECL vs. the nonlesioned controls (Fig. 3F
, H, J).
|
Kainic acid treatment
We considered the possibility that alterations in the expression of HCN1 may have been due to a transient hyperexcitability after ECL. To test whether an increase in neuronal excitability can regulate HCN1 mRNA expression in the adult hippocampus, we used a kainic acid application, which strongly increases neuronal activity in the hippocampus and cortex (41)
. A single subcutaneous (s.c.) injection of kainic acid (12 mg/kg) did not lead to the alteration of HCN1 mRNA expression 6 or 24 h later (data not shown).
Expression of HCN1 protein in neurons
We used two antisera (aq1 and aq2) that were generated against two separate domains in the cytoplasmic tail of the HCN1 protein (28)
. Both antisera gave similar staining patterns. In the unlesioned hippocampus, granule cells showed only a weak staining in their somata and dendrites (Fig. 4
A, D). In the hilus, somata of polymorphic cells were strongly immunopositive (Fig. 4A
, 4E
). CA1 and CA3 pyramidal neurons showed immunopositive staining at their somata and dendrites (Fig. 4A
, 4F
). Two days after ECL, we observed a decrease of immunostaining in the granule cells (Fig. 4B
, 4G
) and hilar neurons (Fig. 4B
, 4H
). CA1 neurons showed a decreased intensity of the immunopositive signal vs. the adult controls (Fig. 4B
, 4I
), corroborating our in situ hybridization data. The translational down-regulation recovered 21 dal as shown by immunostaining (Fig. 4C
).
|
Ih properties and resting physiology of hilar cells in hippocampus of control and EC lesioned animals
To investigate which functional alterations follow from HCN1 decrease, patch-clamp recordings of hilar cells known to express h-currents were performed in the fascia dentata in acute slices. Both HIs and MCs, which possess anomalous rectification (10
, 12
, 42)
, were identified (Fig. 5
A; see Materials and Methods) and recorded to evaluate changes in intrinsic neuronal properties, particularly properties of h-current. Based on the results of the in situ hybridization, we used time groups of 1–3 dal, 5–10 dal, and 30 dal.
Resting membrane properties
The membrane potentials of MCs were not affected by ECL (control: -67±2.5 mV, n=4; 1–10 dal: -62.3±2.9 mV, n=10; 15–30 dal: -61.2±1.1 mV, n=7). The resting input resistance (RN) was increased at 1–10 dal compared with the control (1–10 dal: 477.5±90.1 M vs. control: 212.3±18.3 M, P<0.05; 15–30 dal: 236.4±27.7 M, n=5). The membrane time constant ( 0) was similarly changed (1–10 dal: 65.5±5.0 ms, n=7 vs. control: 48.0±6.4 ms, n=4, P<0.02). There were only minor changes detected in the voltage ‘sag’ (control: 0.81±0.02, n=3, 1–10 dal: 0.89±0.06, n=3, 15–30 dal: 0.73±0.07, n=3).
Control recordings from HIs revealed a VM of -63 ± 1.3 mV (n=8) that only tended to hyperpolarize to -66.1 ± 1.5 mV within 15–30 dal (n=11, n.s.). The RN of HIs increased less markedly and not significantly (1–10 dal: 659.9±102.8 M, n=8, vs. control: 458.4±51.1, n=7). Within 15–30 dal, a significantly smaller RN was measured: 346.3 ± 57.4 M, n = 5 (P<0.05, vs. 1–10 dal). Membrane 0 was similar before and after ECL (41.5±6.1, 48.8±4.2, and 40.0±6.3 during control, 1–10 dal, and 15–30 dal, respectively). In the late postlesion phase (15–30 dal), the voltage ‘sag’ response was enhanced (15–30 dal: 0.61±0.03, n=4 vs. 1–10 dal: 0.79±0.05, n=10, P<0.04., control: 0.76±0.04, n=6).
Properties of hyperpolarization activated currents
A series of voltage-clamp recordings in control animals (n=6) showed the isolation of an h-like current (Fig. 5B
). Voltage steps to -80 mV and more negative (holding potential of -50 mV) revealed inward currents containing two components: an instantaneous fast inward rectifier and a time- and voltage-dependent, noninactivating, inward current for MCs and HIs alike. As shown in many different cell types, the kinetically fast inward rectifier may resemble IKIR and the slower component may derive from Ih (43)
. The relative difference in Ba2+ and Cs+ sensitivity of IKIR and Ih is used as a pharmacological separation between the two hyperpolarization-activated inward currents (35
, 44)
. As illustrated in Fig. 5B
, Islow appeared much less sensitive to the application of 2 mM Ba2+ (at -140 mV: 9.0% block, n=3) than the complete block, which was obtained in the presence of 2 mM Cs+ (-140 mV: 99% block, n=3). Ba2+ did, however, decrease the (instantaneous current) offset by 50%. A full block of Islow under 2 mM Cs+ was also obtained in MC (97%, n=2). This suggests that under our experimental voltage-clamp conditions, the slowly activating inward current pharmacologically resembles Ih. The kinetic properties of the h-current in HIs were similar to previous data obtained from stratum oriens interneurons and HIs (12
, 35
, 36)
. The maximal amplitudes in HIs reached an average of 240 ± 54.5 pA (for -140 mV, n=7), activated with a time constant (1) of 103.0 ± 14.8 ms (n=7). Fitting with a double exponential function was more successful in only four of seven HIs: for these measurements, the fast (FAST) and slow (FAST) activation time constants were 40.3 ± 9.4 and 331 ± 194 ms (n=4), respectively. The voltage for half-maximal (V1/2) activation averaged -86.6 ± 4.0 mV, with a slope factor (k) of 11.7 ± 2.3 (n=7). The reversal potential, estimated using a standard protocol (39)
was -31.8 ± 5.8 mV (n=4), in agreement with previous investigations (35
, 43)
. Extrapolation of the curve revealed a current reversal voltage of -31.8 ± 5.8 mV (n=4), in line with previous investigations (35
, 43)
.
In MCs activation kinetics were characterized by a 1 of 83.3 ± 9.0 ms (n=4), which was not distinct from data of HIs. In MCs, a double exponential fit described the current rise quite well, but only values for FAST could be obtained, probably because of step duration used. The FAST was close to the value obtained with single exponential fitting (64.3±8.0 ms vs. 83.3±9.0, respectively). Basic properties of Ih in MCs were not different from those determined in HIs (MC h-current amplitude: -231.1±23.9 pA, -140 mV, n=4, V 1/2=-82.8±2.1 mV, k=8.7±0.3).
ECL-induced h-current changes
As shown in Fig. 5C-E
, ECL induced a time-dependent change in the kinetics of the h-current in both cell types (the HIs and MCs). A prominent change was found in MCs concerning the rate of current activation that increased approximately twofold (210±34%), i.e., Ih activated more slowly on hyperpolarization (83.3±9.0 ms in control vs. 174.6±31.2 ms during 1–3 dal P<0.003, n=7). With increasing time after ECL, 1 progressively recovered, and at 30 dal the value was indistinguishable from control (82.9±8.7 ms, n=3). Double exponential fits revealed quantitatively comparable changes: FAST was increased by 175 ± 24% (1–3 dal: 112.5±16.0 ms, n=4, vs. control: 64.3±8.0 ms, n=7, P<0.05). At 15 dal, FAST had returned to near control value (66.3±13.5, n=4).
The single exponential time constant of activation of HIs was marginally changed in the early phase after ECL (1–3 dal: 1=128.1±36 ms, n=4; 5–10 dal: 131.8±23 ms, n=7) and not significantly different from the control. In contrast to control HIs, h-currents at 1–3 dal showed no improvement when using double exponential fits (n=6). In the 5–10 dal recovery phase, however, a FAST increase to 210 ± 38% (5–10 dal: 83.8±16 ms, n=5, P<0.05) was observed whereas SLOW (277.5±101.6 ms, n=5) was comparable to the control. At 15 dal, 1 sharply decreased (71.6±9.3 ms, n=5), faster than 5–10 dal (P<0.04), and slightly different from control values (P<0.06). Double exponential fits to h-current activation in HIs during 15 dal indeed revealed faster activation than 5–10 dal (FAST: 53.3±8.0 ms, n=3).
Steady-state voltage activation of Ih
Voltage activation of Ih was analyzed separately for each cell; applying Boltzmann fits to tail current amplitudes (see Materials and Methods), V1/2 and k were compared between groups. Plots of the normalized and averaged data per group were fit with the Boltzmann function and are presented in Fig. 5E
.
For MCs, the V1/2 was shifted directly after the lesion by 17.7 ± 2.0 mV (control: -82.8±2.1, 1–3 dal: -96.2±9.7, P<0.03). Even at 15–30 dal, V1/2 was 13.8 ± 5.4 mV more negative than the control V1/2 (15–30 dal: -96.0±5.4 mV, P<0.07). In all cases, the slope (k) of Ih voltage activation was not affected. For HIs, the initial phase after ECL resulted in a similar change in voltage activation. The V1/2 from 1 to 3 dal was 18.9 ± 4.9 (n=4) hyperpolarized vs. V1/2 in control HIs (1–3 dal: -105.5±4.9 mV, control: -86.6±4.0 mV, P<0.03). Between 5 and 10 dal, the difference was reduced to 5.9 mV, and at 15 dal there was a slight depolarizing shift V1/2 observed of + 5.7 ± 3.1 mV (15 dal: -81.0±3.1 mV, no differences to controls). One month after ECL, V1/2 was shifted by -26.4 ± 3.4 mV (30 dal: -113.0±2.6 mV, n=3, P<0.001 vs. control). The slope (k) was unaffected by ECL.
The reduced Ih activation led to changes in current amplitudes as well. This was most obvious in HIs, and reached only 40% of the control amplitude (-140 mV step, 1–3 dal: -78±22 pA, n=5, control: -240±54.5, P<0.03). Between 1 and 30 dal, there were no detectable changes in whole-cell surface area as determined by the slow-capacitance value (HI control capacitance: 33.7±9.1 pF, n=7, 1–10 dal: 33.1±4.3 pF, n=14).
In MCs, control capacitance values were clearly larger but, similar to HI, not different after ECL (control: 107.3±32.7 pF, 1–10 dal: 142.3±40.5 pF, n=10). As expected, when correcting for the capacitance (pF), a lower h-current density was found at 1–3 dal (-2.6 pA/pF, for the control: -8.6 pA/pF, P<0.05). In MCs, however, the maximal h-current amplitude was still at 81.2 ± 20.7% (n=7) and not significantly different from the control amplitude. ECL did not affect the fast instantaneous (Ba2+-sensitive) current (at -140 mV, corrected for leak: control amplitude -47.3±16.0 pA, n=7, 1–3 dal: -62.8±38 pA, n=4). There was no change in the instantaneous current during any other period (data not shown). In addition, for MCs, the hyperpolarization-activated instantaneous current was not influenced by the ECL (at -140 mV, control: -64.5±23.1 pA, n=4, 1–3 dal: -86.1±32.4, n=7). These data point to specific kinetic and voltage-dependent alterations in Ih properties, leading to a less active Ih. In both MCs and HIs, the early phase after ECL was characterized by slower h-current kinetics and a reduced voltage activation.
Dentate granule cells
In situ hybridization (Fig. 3)
and immunostaining (Fig. 4)
provided evidence for a peculiar expression of HCN1 in GCs. Therefore, we performed voltage-clamp recordings in an attempt to measure HCN-gated currents in GCs of the DG. As shown in Fig. 6
A, hyperpolarizing voltage steps lower than -80 mV revealed steady-state noninactivating inward currents reaching maximal amplitudes of -242.2 ± 31.2 pA (n=13) and activated with 43.4 ± 10.7 ms (n=7) or faster than capacitance settling, i.e., < 10 ms (n=6). Cs+ (2–5 mM) blocked the inward current by 97.6 ± 5.9% (n=7; Fig. 6B
, 6C
). Application of 2 mM of Ba2+ almost completely eliminated the inward current (92.5±2.3%; n=4) (Fig. 6A-C
). These results suggest there is no pharmacological difference of ‘slow’ and ‘fast’ rise time components, which is a characteristic of the h-current (43)
. This was further supported by application of the selective Ih -blocker ZD7288 (45)
, which (at 20 µM) reduced the inward current in only four of seven cells tested by 31.4 ± 11.4% (n=7, Fig. 6B
, 6C
). This block was not accompanied by a blockade of the slow time-dependent component of the initial current amplitude (Fig. 6B
), however, normally found with ZD7288 (cf. ref 32
). The inward currents in granule cells did not change after lesion (at -140 mV, control: -250.8±34.4 pA, n=9, 1–3 dal: -258.4±60.6 pA, n=6, 5–10 dal: -255.4±33.9 pA, n=11).
|
Electron microscopic analysis
We investigated the ultrastructural distribution of HCN1 by electron microscopy of MCs (Fig. 7
A–C) and dentate granule cells (Fig. 7D-F
). The immunoprecipitate was found at postsynaptic sites in the dendrites and spines of both cell types (Fig. 7B
, 7E
), as well as in presynaptic axon-terminals (Fig. 7C
, 7F
). There was a remarkable difference, however, between the two cell types concerning the intracellular localization of the immunoprecipitate in the somata. In MCs, the immunoprecipitate was diffusely scattered in the cytoplasm and also found beneath the cell membrane (Fig. 7A
). In contrast, the immunoprecipitate in granule cells showed a dot-like accumulation in the cytoplasm and was not present near the membrane (Fig. 7D
).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, demonstrate a temporal regulation in fast current activation kinetic and voltage activation, both characteristics in which a predominant role of the HCN1 channel has been implied (36
, 47
48
49)
.
Cellular distribution of HCN1/2/3/4 and the Ih in adult control hippocampus
In this study, we focused on the DG for two reasons: 1) the DG is the first relay station in the hippocampal trisynaptic pathway, which is affected by ECL (2
, 3
, 13)
; and 2) it is in a unique strategic position to modulate the information flow from the EC to the CA regions (8)
. We separately analyzed physiological features of GCs, MCs, and HIs. Whereas GC did not exhibit detectable Ih properties in the somata, the current activation time constant in both HIs and MCs was of rapid kinetics. The fast time constant obtained with double exponential fits was 40 ms for HIs and 60 ms for MCs. Single exponential fits revealed rates of 80–100 ms. The kinetics of current activation were much faster than those of thalamic neurons, but are within the range of other hippocampal cells, such as the CA1 pyramidal neurons or interneurons in the stratum oriens (12
, 35
, 36)
. These fast kinetics suggest a marked contribution of HCN1 channels to inward rectifying currents in these cells (36)
. HCN1 is characterized by a 10–20 mV more positive midpoint in voltage activation kinetics and activation rates 10-fold faster than HCN2 (36
, 48
, 50)
. Our data are in line with the assumption that the HCN1 gated current contributes to 75% of the total inward rectifying h-current at strong hyperpolarized voltage steps (36)
. This dominance is surprising in view of the expression of multiple isoforms of HCN mRNAs hilar neurons. Differences in post-translational modification between the HCN subunits, differences in the stability of certain HCN mRNA, or potential formation of heteromultimeric HCN channels may explain this observation.
Dentate granule cells express the transcripts of all four HCN channels but do not show detectable somatic Ih properties. Our electron microscopic data revealed a peculiar distribution of the immunoprecipitate in the GCs. The dot-like immunoprecipitate was localized only in the cytoplasm, clearly separated from the membrane. This might indicate the presence of the protein but not at the functionally relevant sites within the membrane. This specific intracellular distribution of HCN1 protein might explain both the absence of detectable Ih currents and the weak immunostaining in GCs at the light microscopic level. This is in contrast to MCs, which show a homogeneous distribution of the strong immunosignal in their cell bodies. The lack of somatic Ih in the GCs was not attributable to the quality of cells or slices. The properties of action potentials and membrane time constants indicated healthy cells. Moreover, in the same slices in which Ih was not detectable in GCs, normal Ih was measured in hilar neurons. We suggest that HCN channels are confined to more distal portions of granule cell dendrites that are electrically remote from the soma.
HCN channels and the Ih in the deafferentated hippocampus
The most significant change of the HCN1 mRNA (80% decrease) was observed in the hilus, and the reduced in situ signal expression recovered around 21 dal. Whole-cell recordings of the MCs and HIs revealed that Ih activity was altered in parallel to the altered mRNA expression, and control levels were not attained before 15 dal. The rate of h-current activation was profoundly slower at 1–3 dal and the voltage activation of Ih was characterized by a 20 mV hyperpolarized midpoint of voltage activation in both cell types. As discussed above, the rapid onset in current activation and deactivation and a midpoint of voltage activation in the most positive range have been shown to be mediated by HCN1 channels (36
, 47
, 48
, 50)
. Therefore, the loss of the fast current rise times and a more negative steady-state voltage activation indicate there is in fact a loss of the HCN1-mediated h-current after ECL. This picture may, however, be complicated by the recently described phenotypic heteromerization of HCN1 and HCN2 (48
, 51)
. Whether the heteromerization occurs in MCs or HIs is currently unknown.
The close relationship between membrane conductance (and time constant) and the hyperpolarization-activated current suggests a significant contribution of HCN1 to the resting conductance of HIs and MCs, as has also been reported by others on different neuronal cell types (35
, 50)
. However, the decrease in membrane conductance of both cell types after lesion was not accompanied by a hyperpolarization of the membrane (35)
, as would be expected from a reduction of a depolarizing current like Ih. Probably a concomitant decline of outward potassium currents precludes a detection of the anticipated hyperpolarization (52)
. It is unclear at this time why entorhino-hippocampal disconnection leads specifically to a reduction of HCN1 whereas HCN2/3 and 4 are not altered.
Functional consequences of HCN1/Ih changes after ECL
During the early stages after ECL, rats exhibit spontaneous recurrent limbic seizures (53
, 54)
. It is tempting to speculate whether the transient decrease in HCN1 expression and Ih densities relate to these changes. Ih-type currents contribute to the pacemaker depolarization that generates oscillatory activity (55
, 56)
. It has been suggested that Ih also ensures sustained repetitive firing (35)
and augments dendritic attenuation (57)
. However, an attenuation of the hyperpolarization-activated current reduces synchronization of the hilar network (56
, 58)
.
The main projection neurons in the DG, the GCs, did not exhibit somatic Ih properties in our study, indicating that the functional consequences of lesion-induced changes of HCN1/Ih properties should be attributed to cells located in the hilus. In fact, MCs are the main cells of origin of the commissural/associational projection to the contralateral DG. This projection is glutamatergic and thus is in a position to modulate GC excitability (59)
. Moreover, MCs are primarily activated by granule cell axons and function as an activator of GABAergic inhibitory neurons (60
, 61)
. A decrease in Ih after ECL, as revealed in our study, might well cause a decrease in activation of inhibitory neurons and, in turn, reduce inhibition of GCs. This assumption would be in line with the hypothesis predicting that such changes should result in a hyperexcitability of the DG (62)
. The heterogeneous population of HIs, on the other hand, is mostly inhibitory, and these cells innervate the outer molecular layer of the fascia dentata and pyramidal neurons in CA1-CA3 (9
, 63
, 64)
. Thus, this intrinsic inhibitory system is responsible for the hyperpolarizing inhibitory postsynaptic potential of pyramidal cells of CA1-CA3 and is capable of effectively blocking transmission from GCs to CA3 (65
, 66)
. Our data demonstrate a down-regulation of the HCN1 mRNA and protein in these cells that could result in a decrease in Ih after lesion, finally leading to a disinhibition of the GCs. Our data indicate the involvement of cell type specific changes of distinct physiological properties, which might explain the imbalance of inhibition and excitation in the DG after ECL (19)
.
The dentate spikes are an important physiological feature of the hilar region consisting of large field potentials (11)
. These spikes are preceded by a decrease in hilar interneuronal firing (11
, 67)
. In synchrony with the dentate spikes in the hilus, giant IPSPs in the CA3 region and depolarizing potentials at dentate GCs occur. The bilateral removal of entorhinal input causes a decrease in dentate spikes of the hilar region in vivo (11)
. A reduction of dentate spikes was still observed even after 3 wk postlesion and returned to normal levels in rats with a partial ECL within 2 wk (11)
. The initial reduction of Ih after ECL might exert ‘beneficial effects‘ through hyperpolarization of neurons (50
, 68)
or decrease in the maximal firing frequency (35)
or a ‘deleterious effect‘ through an increased effective electrotonic spread, thus augmenting excitatory input. In both ways, our data demonstrating a transient decrease of Ih after unilateral lesion are in line with these in vivo data.
At later stages, septohippocampal cholinergic projections (69
, 70)
and glutamatergic fibers originating in the contralateral EC (7)
sprout and form new excitatory synapses with the remaining dendrites. In addition, commissural GABAergic fibers, which also terminate in the outer two-thirds of the molecular layer (14)
, show an increase in axon terminals and synaptic contacts (17)
. Thus, the initial reduction of the excitatory input after ECL is partially compensated for by the successive expansion of cholinergic and glutamatergic fibers. In this temporal sequence, we could also observe a recovered HCN1 expression and Ih current. This indicates that the recovered Ih current participates in the process of the formation of new synaptic contacts by sprouting axons after ECL.
Along with previous studies, the results reported here suggest a dynamic alteration in the expression of HCN1 in the lesioned hippocampus that leads to changes of the Ih current in the system. Furthermore, the depression of Ih is not permanent and coincides with the initial degeneration after sprouting processes, resulting in a newly adjusted excitatory/inhibitory balance.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication April 16, 2001.
Revision received August 20, 2010.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
