Professional Documents
Culture Documents
Epilepsy Research
journal homepage: www.elsevier.com/locate/epilepsyres
A R T I C LE I N FO A B S T R A C T
Keywords: Medial temporal lobe epilepsy (mTLE)–the most common form of focal epilepsy–is defined by recurrent partial
Carbamazepine seizures originating within the medial temporal lobe. Such seizures are commonly associated with the anterior
Dorsal hippocampus hippocampus (as opposed to the posterior hippocampus), and refractory to the currently available anti-epileptic
Intrinsic excitability drugs (AED) for about one third of patients. Unfortunately, the mechanisms driving seizure generation and AED
Anti-Epileptic drug
efficacy along the longitudinal hippocampal axis remain poorly understood. Recently, several groups in-
CA1 pyramidal neuron
vestigating differences in excitability along the rodent longitudinal hippocampal axis have demonstrated that
CA1 pyramidal neurons from the rodent ventral hippocampus (the rodent homolog of the human anterior
hippocampus) are intrinsically more excitable than their dorsal counterparts (the rodent homolog of the human
posterior hippocampus). This phenotypic difference is accompanied by significant differences in gene expression
along the longitudinal hippocampal axis, which include gene products–such as voltage-gated sodium channel β-
subunits–known to influence AED efficacy. Given this phenotypic heterogeneity, and the differential expression
of gene products known to influence anti-epileptic drug efficacy, we sought to investigate the efficacy of the
classical use-dependent sodium channel blocker, carbamazepine, in CA1 pyramidal neurons across the long-
itudinal hippocampal axis. Accordingly, we performed whole-cell current-clamp recordings on CA1 pyramidal
neurons from acute hippocampal slices prepared from the dorsal and ventral hippocampus, and found that acute
exposure to 100 μM carbamazepine induced a significantly greater suppression of repetitive firing for dorsal
neurons relative to ventral neurons by inducing profound spike frequency adaptation (SFA). Moreover, we
observed a small, but significant depolarization of resting membrane potential (RMP) for dorsal neurons (but not
ventral neurons), following exposure to carbamazepine. Together, these observations demonstrate that carba-
mazepine’s effect is concentrated in the dorsal hippocampus, which could provide meaningful insight into the
side effect profile of carbamazepine (and related anti-epileptic drugs) in non-epileptic tissue, and inform future
work investigating the mechanisms of carbamazepine resistance in epileptic tissue.
Abbreviations: mTLE, medial temporal lobe epilepsy; AED, anti-epileptic drug; DHC, dorsal hippocampus; dIHC, dorsal intermediate hippocampus; vIHC, ventral intermediate hip-
pocampus; VHC, ventral hippocampus; ISA, somatodendritic A-type current; IM, m-current; GIRK, channel, G-protein coupled inwardly recitifying potassium channel; CBZ, carbamazepine;
aCSF, artificial cerebrospinal fluid; DNQX, 6,7-dinitroquinoxaline 2,3-dione; DMSO, dimethyl sulfoxide; Rin, input resistance; τM, membrane time constant; Vthr, threshold voltage; APPeak,
action potential peak; APsize, action potential size; APhw, action potential half width; ISI, interspike interval; SFA, spike frequency adaptation; PTSD, post-traumatic stress disorder; FPKM,
fragments per kilobase of exon per million fragments mapped
⁎
Corresponding author at: 2000 N Parkway, Memphis, TN, 38112, USA.
E-mail address: doughertyk@rhodes.edu (K.A. Dougherty).
https://doi.org/10.1016/j.eplepsyres.2018.05.014
Received 7 March 2018; Received in revised form 11 May 2018; Accepted 29 May 2018
Available online 09 June 2018
0920-1211/ © 2018 Elsevier B.V. All rights reserved.
M.C. Evans, K.A. Dougherty Epilepsy Research 145 (2018) 63–72
64
M.C. Evans, K.A. Dougherty Epilepsy Research 145 (2018) 63–72
Instruments, San Fransisco, CA, USA), and filled with an internal so- 2.5. Statistical evaluation
lution with the following composition (in mM): 120 potassium gluco-
nate, 20 KCl, 10 HEPES, 4 NaCl, 4 Mg-ATP, 0.3 Na-GTP, and 7 (Tris)2- Statistical significance between paired groups was determined using
phosphocreatine (pH 7.3, adjusted with KOH; 300 mOsm, adjusted with a Paired Student’s t test when n ≥ 7, and a Wilcoxon Signed-Rank test
sucrose). Whole-cell current-clamp recordings were performed using a when n < 7, the data were not normally distributed, or the variances
Dagan BVC-700A amplifier (Dagan, Minneapolis, MN, USA) under the were unequal. Statistical significance between independent groups was
control of a MacMini computer (Apple Computer Company, Cupertino, determined using Student’s t test when n ≥ 7, and a Wilcoxon Ranked-
CA, USA) through an ITC-18 computer interface (InstruTech, Port Sum test when n < 7, the data were not normally distributed, or the
Washington, NY, USA). Signals from were low-pass filtered at 3 kHz and variances were unequal. Differences between firing frequency versus
sampled at 10 kHz for subthreshold experiments, whereas signals from current injection (F–I) relationships were determined using a Two-Way
suprathreshold experiments were low-pass filtered at 5 kHz and sam- Repeated Measures ANOVA test, followed by a Bonferroni correction
pled at 40 kHz. Neurons were allowed to stabilize for at least five for multiple comparisons. Paired F–I relationships were evaluated using
minutes after break in, and the pipette capacitance and access re- repeated measures in two dimensions (time and current injection),
sistance were compensated throughout the experiment prior to each whereas unpaired F–I relationships were evaluated using repeated
recording. Experiments were terminated if the series resistance ex- measures in one dimension (current injection). Spearman’s Rank
ceeded 30 MΩ, and small DC current injections were often used to Correlation test was used to evaluate correlation between datasets.
maintain the membrane potential at −65 mV. Membrane potentials Unless otherwise stated, statistical analyses and graphical displays were
have not been adjusted to account for a liquid junction potential of accomplished using IGOR Pro v6.35 (Wavemetrics, Lake Oswego, OR,
approximately 8 mV. Neurobiotin (0.1–0.2%; Vector Laboratories, USA). Two-Way ANOVA tests and correlation analyses were performed
Burlingame, CA) was added to the internal solution in order to facilitate using Prism 7 (GraphPad Software, La Jolla, CA, USA). All hypothesis
the subsequent visualization of individual neurons and to verify intact testing was conducted with α = 0.05, and the specific statistical tests
apical and basal dendritic trees. After each experiment, slices were fixed used to evaluate each dataset are indicated in the corresponding figure
in 0.1 M phosphate buffer with 3% gluteraldehyde, stored at 4 °C for at legends. All data are presented as SEM, unless otherwise stated.
least 48 h, and processed using an avidin-HRP system activated by
diaminobenzidine (DAB; Vector Laboratories). Processed slices were 3. Results
mounted in glycerol and visualized using a compound microscope with
a 10X objective. 3.1. Baseline parameters
2.3. Drugs and solutions In order to evaluate the influence of carbamazepine along the
longitudinal hippocampal axis, we performed whole-cell current-clamp
All experiments were performed in an aCSF supplemented with recordings on CA1 pyramidal neurons from DHC and VHC slices pre-
20 μM 6,7-dinitroquinoxaline 2,3-dione (DNQX; Abcam, Cambridge, pared from adult male rats (4.7 ± 0.5 months old; 403 ± 12 g). DHC
MA, USA) and 2 μM Gabazine (Abcam) to block AMPA receptors and slices were sampled from a range of locations along the longitudinal
GABAA receptors, respectively, and thereby suppress spontaneous ac- hippocampal axis that included both the dorsal hippocampus and the
tivity within the slice. Carbamazepine (Sigma-Aldrich, St Louis, MO, dorsal intermediate hippocampus (dIHC; as defined by Malik et al.,
USA) was dissolved in DMSO (Sigma-Aldrich) to yield a 100 mM stock 2016), whereas VHC slices were taken mostly from the ventral inter-
solution, which was subsequently aliquoted and frozen until needed. mediate hippocampus (vIHC), rather than the ventral hippocampus
Carbamazepine aliquots were diluted 1000-fold into aCSF to yield near the temporal pole (Malik et al., 2016). CA1 pyramidal neurons
100 μM carbamazepine and 0.1% DMSO. This CBZ concentration is from this segment of the longitudinal hippocampal axis (DHC–dIHC–-
approximately 2–3X the plasma concentration required to suppress vIHC) are generally similar with regard to their morphology and
hind limb extension in the maximal electroshock seizure test in rats baseline intrinsic parameters, such as Rin (at −65 mV), τm (at −65 mV),
(Bialer et al., 2004). DMSO was diluted 1000-fold in aCSF to yield 0.1% and Vthr (at −65 mV) (Malik et al., 2016). This observation stands in
DMSO solutions for vehicle control experiments. contrast to the region of the VHC closest to the temporal pole, where
CA1 pyramidal neurons exhibit a significantly different morphological
2.4. Data analysis and electrophysiological phenotype (Malik et al., 2016; Dougherty
et al., 2012). Therefore, we limited our study to the DHC–dIHC–vIHC
Suprathreshold parameters, such as the threshold voltage (Vthr), segment of the longitudinal hippocampal axis, and will hereafter refer
maximal dV/dt, and action potential peak (APpeak), were determined to the vIHC as the VHC. Accordingly, representative DHC and VHC
through phase–plane analysis, where Vthr was defined as the membrane slices, with corresponding representative filled CA1 pyramidal neurons
potential were the dV/dt first exceeded 20 mV/ms. The action potential are displayed in Fig. 1A–B. Baseline parameters such as RMP, Rin (at
size was defined as the difference between the APpeak and Vthr, and the −65 mV), Rin (at RMP), and τm (at −65 mV) were not significantly
action potential half-width was defined as the width of the action po- different between DHC and VHC neurons (DHC RMP = −63.4 ±
tential waveform at one half of the action potential size. The spike 0.9 mV, n = 15, Rin (at −65 mV) = 58.0 ± 2.5 MΩ, n = 15, Rin (at
frequency adaptation (SFA) ratio was determined from action potential RMP) = 59.6 ± 3.9 MΩ, n = 11, τm (at –65 mV) = 20.0 ± 1.1 ms,
trains elicited by one second long current injections, where the current n = 10; VHC RMP = −62.0 ± 0.8 mV, n = 15, Rin (at
magnitude was varied to produce trains containing 10–12 action po- −65 mV) = 61.8 ± 3.4 MΩ, n = 15, Rin (at RMP) = 71.5 ± 8.9 MΩ,
tentials. The SFA was then calculated by dividing the duration of the n = 11, τm (at –65 mV) = 21.0 ± 1.9, n = 10; Fig. 1C–E). These values
final interspike interval (ISI) by the duration of the first ISI. The input were similar to those reported in a previous study using adult rats under
resistance (Rin) was determined as the slope of the voltage–current plot similar conditions (i.e., without blocking NMDA receptors), although
constructed from the steady-state voltage responses to a family of the Rin values for VHC neurons were slightly lower than previously
current injections ranging from −50 to 50 pA (only the linear region of reported (Clemens and Johnston, 2014). This is likely due to the
these plots were used for determining the Rin). The membrane time longitudinal location of our VHC slices (see above). The voltage
constant (τm) was determined by averaging the voltage responses to threshold (Vthr) of single action potentials elicited by current injections
100, 1 ms square current injections (100 pA), and fitting the average from −65 mV were also found to be indistinguishable between DHC
response with a double exponential curve. The slower of the two time and VHC neurons in both the 10 ms or 50 ms timeframes (Fig. 1F; DHC
constants was taken as τm. Vthr (10 ms) = −49.2 ± 1.0 mV, n = 13, Vthr (50 ms) = −47.0 ±
65
M.C. Evans, K.A. Dougherty Epilepsy Research 145 (2018) 63–72
Fig. 1. Baseline properties of CA1 pyramidal neurons across the longitudinal hippocampal axis of adult rats. A–B) Representative slices (left) and filled CA1
pyramidal neurons (right) from the dorsal (A) and ventral (B) hippocampus. C–E) The resting membrane potential (C), Rin (at −65 mV) (D; circles), the Rin (at RMP)
(D; squares), and the τm (at −65 mV) (E) of DHC (blue) and VHC neurons (black) were not significantly different (Student’s t-test, p > 0.05). F) The voltage threshold
measured 10 ms (circles) and 50 ms (squares) after the onset of the current stimulus were not significantly different between DHC (blue) and VHC (black) neurons
(Student’s t-test, p > 0.05). The voltage threshold for VHC neurons was significantly more depolarized in the 50 ms timeframe than in the 10 ms timeframe (Students’s
t-test, p < 0.05), whereas the voltage threshold was not significantly different for DHC neurons between these timeframes (Student’s t-test, p > 0.05). Average
values are represented as horizontal lines. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
0.8 mV, n = 13; VHC Vthr (10 ms) −49.6 ± 0.6 mV, n = 12, Vthr firing frequency of both DHC neurons and VHC neurons, although VHC
(50 ms) = −46.5 ± 0.8 mV, n = 12). Although some subtle gradients neurons continued to fire throughout the current stimulus, whereas
in intrinsic excitability are likely present, these observations demon- DHC neurons often fired a few action potentials at the beginning of the
strate that several key factors of intrinsic excitability are generally si- current stimulus before ceasing firing altogether (Fig. 3A–C). As in the
milar along a significant portion of the longitudinal hippocamal axis. shorter, one second timeframe, 100 μM CBZ induced a significantly
This overall similarity serves as a foundation for investigating the in- greater suppression in the F–I relationships for DHC neurons than their
fluence of AEDs across this region of the longitudinal hippocampal axis. VHC counterparts, despite indistinguishable baseline F–I relationships
(Fig. 3D–E). There was no significant reduction of the F–I relationship
in 0.1% DMSO control experiments (Supplementary Fig. 2). Moreover,
3.2. Carbamazepine-induced suppression of repetitive firing is greater in the firing frequency in the presence of 0.1% DMSO increased over time,
DHC neurons than VHC neurons yielding a fold change in firing frequency (10 s, 250 pA current injec-
tions) greater than one for both DHC and VHC neurons (Fig. 4; DHC
We evaluated the effect of carbamazepine on repetitive firing in Fold Change (DMSO) = 1.63 ± 0.16, n = 6; VHC Fold Change
DHC and VHC neurons before, and 20–60 min after applying 100 μM (DMSO) = 1.37 ± 0.18, n = 5). When compared to 0.1% DMSO con-
CBZ through the bath solution (solution exchange was conducted at trol experiments, 100 μM CBZ significantly reduced the firing frequency
RMP, which was allowed to float throughout the wash-in period). of both DHC and VHC neurons (Fig. 4; DHC Fold Change
100 μM CBZ significantly reduced the firing frequency for action po- (CBZ) = 0.13 ± 0.06, n = 8; VHC Fold Change (CBZ) = 0.65 ± 0.13,
tential trains elicited by one-second long current injections from n = 7). The fold change in firing frequency in response to 100 μM CBZ
−65 mV (Fig. 2A–C). This effect was especially pronounced in DHC was significantly lower for DHC neurons than their VHC counterparts
neurons, which often ceased firing after just a few action potentials (Fig. 4).
(Fig. 2A, left column). In contrast, VHC neurons tended to fire
throughout the current injection, albeit at a lower frequency (Fig. 2A,
right column). Although the F–I baseline relationships for DHC and 3.3. Carbamazepine influences subthreshold and supratheshold properties
VHC neurons were indistinguishable (Fig. 2D), the F–I relationship for of CA1 pyramidal neurons from the dorsal hippocampus
DHC neurons was significantly lower than their VHC counterparts in
the presence of 100 μM CBZ (Fig. 2E). This differential suppression of In order to understand the mechanism(s) driving the dispropor-
the F–I relationship was not observed in vehicle control experiments, tionate effect of CBZ on DHC neurons, we measured carbamazepine’s
where only 0.1% DMSO was applied to DHC and VHC neurons (Sup- effect on single action potentials in DHC and VHC neurons. Single ac-
plementary Fig. 1). tion potentials were elicited by brief (10 ms) current injections of
We next extended the duration of the current injection from one variable amplitude, such that the first action potential occurred within
second to ten seconds in order to examine the influence of carbama- a ± 1 ms time window centered 8 ms following the onset of the current
zepine on firing for DHC and VHC neurons throughout prolonged action stimulus before and after 100 μM CBZ wash in (Fig. 5A; Table 2). This
potential trains. Accordingly, both DHC and VHC neurons fired short timeframe was chosen to minimize the influence of slowly acti-
throughout the 10 s, 250 pA current injection in the absence of carba- vating and inactivating ion channels. Under these conditions, baseline
mazepine (Fig. 3A). Exposure to 100 μM CBZ significantly reduced the parameters from DHC and VHC neurons were indistinguishable
66
M.C. Evans, K.A. Dougherty Epilepsy Research 145 (2018) 63–72
Fig. 2. Carbamazepine-induced suppression of repetitive firing in response to short current injections. A) Representative traces from DHC (left) and VHC
(right) neurons elicited by 1 s, 330 pA current injections from −65 mV before (upper traces) and after (lower traces) exposure to 100 μM carbamazepine. B–C) F–I
relationships for DHC (B) and VHC (C) neurons before (hollow circles) and after (filled circles) exposure to 100 μM carbamazepine. Carbamazepine significantly
suppressed repetitive firing for DHC neurons and VHC neurons (Two-way repeated measures ANOVA; Bonferroni test for multiple comparisons, p < 0.05). D–E) F–I
relationships of DHC and VHC neurons before (D) and after (E) exposure to 100 μM carbamazepine. In the presence of 100 μM CBZ, the F–I relationship was
significantly lower for DHC neurons than VHC neurons (Two-way repeated measures ANOVA; Bonferroni test for multiple comparisons, p < 0.05).
(Table 2), except for the maximum dV/dt, which was significantly n = 8; VHC ΔVthr = −0.6 ± 0.7 mV, n = 6), action potential size
lower for VHC neurons than DHC neurons (DHC max dV/ (DHC ΔAPsize = −3.5 ± 1.6 mV, n = 8; VHC ΔAPsize = −3.8 ±
dt = 437.1 ± 16.5 mV/ms, n = 13; VHC max dV/dt = 371.2 ± 2.0 mV, n = 6), and action potential halfwidth (DHC ΔAPhw =
12.8 mV/ms, n = 12) Subsequent exposure to 100 μM CBZ did not 0.02 ± 0.02 ms, n = 8; VHC ΔAPhw = 0.04 ± 0.03 ms, n = 6) for ei-
significantly alter the voltage threshold (DHC ΔVthr = −1.2 ± 2.0 mV, ther DHC or VHC neurons (Fig. 5; Table 2). However, the maximal dV/
Fig. 3. Carbamazepine-induced suppression of repetitive firing in response to long current injections. A) Representative traces from DHC (left) and VHC
(right) neurons elicited by 10 s, 250 pA current injections from −65 mV before (upper traces) and after (lower traces) exposure to 100 μM carbamazepine. B–C) F–I
relationships for DHC (B) and VHC (C) neurons before (hollow circles) and after (filled circles) exposure to 100 μM carbamazepine. Carbamazepine significantly
suppressed repetitive firing for DHC neurons and VHC neurons (Two-way repeated measures ANOVA; Bonferroni test for multiple comparisons, p < 0.05). D–E) F–I
relationships of DHC and VHC neurons before (D) and after (E) exposure to 100 μM carbamazepine. In the presence of 100 μM CBZ, the F–I relationship was
significantly lower for DHC neurons than VHC neurons (Two-way repeated measures ANOVA; Bonferroni test for multiple comparisons, P < 0.05).
67
M.C. Evans, K.A. Dougherty Epilepsy Research 145 (2018) 63–72
dt was significantly reduced for both DHC and VHC neurons in the
presence of 100 μM CBZ (Fig. 5D; DHC Δmax dV/dt = −41.6 ±
13.0 mV/ms, n = 8; VHC Δmax dV/dt = −27.7 ± 7.2 mV/ms, n = 6).
Despite this effect, single action potentials were mostly unaffected by
the presence of 100 μM CBZ, which is consistent with carbamazepine’s
classical mechanism of action as a use-dependent sodium channel
blocker (Ragsdale et al., 1991).
Although single action potentials were mostly unaffected, carba-
mazepine had a profound effect on action potential trains for both DHC
and VHC neurons (Figs. 2 and 3). Accordingly, we examined carba-
mazepine’s influence on spike frequency adaptation, by calculating the
SFA ratio from action potential trains containing 10–12 action poten-
tials (Fig. 6A). The baseline SFA ratio was significantly larger for DHC
neurons compared to VHC neurons (Fig. 6B; DHC SFA ratio = 2.95 ±
0.31, n = 11; VHC SFA ratio = 1.39 ± 0.19, n = 10), which is con-
sistent with previous reports (Malik et al., 2016). However, in the
presence of 100 μM CBZ, it became impossible to elicit the 10–12 action
potentials required to determine the SFA ratio in all but two DHC
Fig. 4. Summary of carbamazepine’s effect on firing frequency. The fold neurons–the SFA ratio for these two neurons increased 2.41 and 1.67
change in firing frequency (relative to the baseline firing frequency) for action fold. Alternatively, 100 μM CBZ did not significantly alter the SFA ratio
potential trains elicited by 10 s, 250 pA current injections, is significantly lower for VHC neurons (Fig. 6B; VHC ΔSFA ratio = 1.59 ± 0.13, n = 5), and
in the presence of 100 μM carbamazepine for both DHC and VHC neurons re- no significant effect was observed in 0.1% DMSO control experiments
lative to 0.1% DMSO controls (Wilcoxon’s Ranked Sum Test, p < 0.05). The (Fig. 6B; DHC ΔSFA ratio = 0.72 ± 0.06, n = 5; VHC ΔSFA
fold change in firing frequency is significantly lower for DHC neurons than VHC
ratio = 0.88 ± 0.24, n = 4). These observations demonstrate that
neurons following 100 μM CBZ exposure (Wilcoxon’s Ranked Sum Test,
carbamazepine induces a profound spike frequency adaptation in DHC
p < 0.05), whereas there was no difference in the fold change in firing fre-
quency following 0.1% DMSO exposure (Wilcoxon’s Ranked Sum Test,
neurons, but spares their VHC counterparts.
p > 0.05). Carbamazepine’s effect was not limited to supratheshold properties.
Surprisingly, carbamazepine influenced the subthreshold properties of
Fig. 5. Carbamazepine does not significantly alter single action potentials. A) Single action potentials elicited by 10 ms current injections before and after
exposure to 100 μM CBZ in DHC (upper traces) and VHC (lower traces) neurons. B–C) The threshold voltage (B) and action potential size (C) were not significantly
different following 100 μM CBZ exposure for DHC (blue circles; Paired Student’s t-test, p > 0.05) or VHC (black circles; Wilcoxon’s Signed Rank test, p > 0.05)
neurons. D) Carbamazepine significantly reduced the max dV/dt for both DHC (Paired Student’s t-test, p < 0.05) and VHC neurons (Wilcoxon’s Signed Rank test,
p < 0.05). E) The action potential halfwidth was not significantly different following carbamazepine exposure for either DHC (Paired Student’s t-test, p > 0.05) or
VHC neurons (Wilcoxon’s Signed Rank test, p > 0.05). Average values are represented as horizontal lines. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)
68
M.C. Evans, K.A. Dougherty Epilepsy Research 145 (2018) 63–72
Table 2
Single action potential parameters.
Voltage Threshold (mV) Maximum dV/dt (mV/ms) AP size (mV) AP halfwidth (ms)
The parameters of single action potentials elicited approximately 8 ms after the onset of a square pulse from −65 mV are displayed before (baseline), and after
exposure to either 100 μM CBZ or 0.1% DMSO. Statistical significance was determined using either a paired Student’s t test (n ≥ 7) or a Wilcoxon’s Signed Rank test
(n < 7), with p < 0.05. Significant differences are indicated by superscripted letters, and n values are indicated in parentheses.
Fig. 6. Carbamazepine-induced enhancement of spike frequency adaptation across the longitudinal hippocampal axis. A) The SFA ratio was determined by
dividing the duration of the final ISI (dotted lines) by the first ISI (solid lines) for one second long action potential trains containing 10–12 action potentials. B) The
baseline SFA ratio was significantly higher for DHC neurons (blue symbols) than VHC neurons (black symbols) for both 0.1% DMSO control (squares; Wilcoxon’s
Ranked Sum test, p < 0.05) and 100 μM CBZ experiments (circles; Wilcoxon’s Ranked Sum test, p < 0.05). Exposure to 100 μM CBZ induced a profound spike
frequency adaptation in DHC neurons that made it impossible to elicit 10–12 action potentials in the one second time frame in all but two neurons, whereas VHC
neurons continued to fire throughout the one second interval, although 100 μM CBZ did not significantly alter the SFA ratio (Wilcoxon’s Signed Rank test, p > 0.05).
(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
DHC neurons, as 100 μM CBZ induced a small, but significant depo- 4. Discussion
larization of RMP for DHC neurons, but not VHC neurons (Fig. 7A; DHC
ΔRMP = 3.0 ± 1.1 mV, n = 9; VHC ΔRMP = −0.2 ± 0.6 mV, n = 8). 4.1. Summary
However, the presence of 100 μM CBZ did not significantly alter the Rin
(at − 65 mV) (Fig. 7B–C; DHC ΔRin (at -65 mV) = -0.8 ± 5.6 MΩ, In this study, we investigated the influence of carbamazepine on CA1
n = 6; VHC ΔRin (at −65 mV) = 10.1 ± 9.7 MΩ, n = 7), and no pyramidal neurons across the longitudinal hippocampal axis, and have
changes in either the RMP nor the Rin (at −65 mV) were observed found that carbamazepine 1) significantly suppresses repetitive firing in
during 0.1% DMSO control experiments (DHC ΔRMP = 1.3 ± 0.8 mV, DHC neurons through an enhancement of spike frequency adaptation
ΔRin (at -65 mV) = 7.6 ± 7.2 MΩ, n = 6; VHC ΔRMP = −1.3 ± (Figs. 2–4 and 6), 2) induces a slight depolarization of the RMP in DHC
0.8 mV, ΔRin (at −65 mV) = 5.1 ± 8.6 MΩ, n = 7). Furthermore, neurons (Fig. 7), and 3), that these two effects were significantly corre-
100 μM CBZ did not alter τm (at −65 mV) for DHC neurons (Fig. 7D–E; lated across the longitudinal hippocampal axis (Fig. 8). Carbamazepine
Δτm (at −65 mV) = 0.8 ± 3.3 ms, n = 6), but induced a small, but also suppressed repetitive firing in VHC neurons, but this effect was
significant increase in the τm (at −65 mV) for VHC neurons (Fig. 7D–E; significantly smaller than their DHC counterparts (Figs. 2–4). Moreover,
VHC Δτm (at −65 mV) = 3.1 ± 2.0 ms, n = 6). No significant differ- no change in the RMP was observed following 100 μM CBZ exposure for
ences were observed for τm (at −65 mV) following exposure to 0.1% VHC neurons (Fig. 7A). Together, these observations suggest a unique
DMSO (DHC Δτm (at −65 mV) = 5.9 ± 2.1, n = 4; VHC Δτm (at sensitivity to carbamazepine for DHC neurons relative to their VHC
−65 mV) = 2.1 ± 0.8 ms, n = 4). Interestingly, cells that depolarized counterparts. The potential mechanisms and physiological implications
in the presence of 100 μM CBZ tended to show a greater suppression of of these findings are further discussed below.
their F–I relationships, as the ΔRMP was significantly correlated with
the fold change in firing frequency along the longitudinal hippocampal 4.2. Comparison with previous work
axis (Fig. 8A). Neither the suprathreshold nor the subthreshld effect of
carbamazepine significantly correlated with the original RMP, at which Interestingly, this regional difference in carbamazepine sensitivity
100 μM CBZ was applied to the slice (data not shown). Together, these emerged from a population of CA1 pyramidal neurons with similar
observations demonstrate that carbamazepine exerts a unique influence baseline intrinsic properties (e.g., RMP, Rin (at −65 mV), Rin (at RMP),
over the intrinsic excitability of CA1 pyramidal neurons from the DHC. τm (at −65 mV), and Vthr; Fig. 1). At first glance, this baseline similarity
between DHC and VHC neurons seemingly contradicts the significant
differences in intrinsic excitability reported across the longitudinal
69
M.C. Evans, K.A. Dougherty Epilepsy Research 145 (2018) 63–72
Fig. 7. 100 μM carbamazepine influences subthreshold properties of CA1 pyramidal neurons. A) The RMP of DHC neurons (blue circles), but not VHC neurons
(black circles) was significantly more depolarized following exposure to 100 μM CBZ (Paired Student’s t-test, p < 0.05). B–C) the Rin (at −65 mV) was not sig-
nificantly different following exposure to 100 μM CBZ for either DHC (B blue circles; C blue traces), or VHC neurons (B black circles; C black traces; Paired Student’s t-
test, p > 0.05). D–E) the τm (at −65 mV) was not significantly different following exposure to 100 μM CBZ for DHC neurons (D blue circles; E left, blue/gray traces;
Wilcoxon’s Signed Rank test, p > 0.05) but showed a small, yet significant increase for VHC neurons (D black circles; E right, black/gray traces; Wilcoxon’s Sign Rank
test, p < 0.05). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
for this study were significantly older than those used in previous stu-
dies (Dougherty et al., 2012, 2013; Malik et al., 2016; Kim and
Johnston, 2015), 2) NMDA receptor blockers were omitted from our
cocktail of synaptic blockers, and 3) slices for this study were prepared
using a slightly different blocking strategy than previous studies, which
doesn’t sample the extreme ventral hippocampus near the temporal
pole. Recent work by Clemens and Johnston (2014), using adult rats
and a similar synaptic blocker cocktail (i.e., without NMDA blockers)
reported an ∼3 mV difference between the average RMPs, and an ∼
20 MΩ difference in the Rin (at −65 mV) of DHC and VHC neurons
(Clemens and Johnston, 2014). Our baseline RMP and Rin (at −65 mV)
values matched those previously reported by Clemens and Johnston
(2014) for DHC neurons, whereas our Rin (at −65 mV) was lower than
previously reported for VHC neurons (62 MΩ versus ∼80 MΩ from
Clemens and Johnston (2014)). This discrepancy likely stems from the
relative longitudinal position of the slices used for this study. Accord-
Fig. 8. Correlation between carbamazepine’s subthreshold and supra-
ingly, Malik et al. (2016) have recently reported tends in intrinsic ex-
threshold effects across the longitudinal hippocampal axis. DHC (blue
circles) and VHC neurons (black circles) were combined in order to evaluate the citability along the longitudinal axis of young adult rats, wherein the
correlation between the ΔRMP with the fold change in firing frequency for CA1 Rin (at −65 mV) remains constant around 60 MΩ along the dorsal
pyramidal neurons across the longitudinal hippocampal axis. This relationship portion of the longitudinal axis until jumping to about 80 MΩ near the
was described assuming a linear function with the following parameters: y- temporal pole (Malik et al., 2016). The slices used in this study were
intercept = 3.2 mV, slope = −4.1 mV/fold change, and correlation was eval- prepared mostly from the vIHC, rather than the extreme VHC, which
uated using Spearman’s Rank Correlation test (R = −0.54, p = < 0.05, n = 15). places our observations in good agreement with those previously re-
(For interpretation of the references to colour in this figure legend, the reader is ported by Malik et al. (2016). This reasoning also accounts for the si-
referred to the web version of this article.) milarity in Vthr that we observed across the longitudinal hippocampal
axis, as Vthr remains similar across the DHC, dIHC, and vIHC, before
hippocampal axis (Dougherty et al., 2012; Hönigsperger et al., 2015; depolarizing in the extreme VHC (Malik et al., 2016). The similarity in
Malik et al., 2016; Kim and Johnston, 2015; Milior et al., 2016). RIn (at −65 mV) and Vthr along this region of the longitudinal hippo-
However, our current experimental conditions differ from those re- campal axis spares our interpretation from the potentially confounding
ported in previous studies in several important ways: 1) The rats used influence of differences in baseline intrinsic excitability.
70
M.C. Evans, K.A. Dougherty Epilepsy Research 145 (2018) 63–72
4.3. Potential mechanisms VHC; Figs. 7 and 8). Interestingly, the depolarization of RMP occurred
in DHC neurons, rather than in VHC neurons, where a depolarization of
The main finding of this work is the region-specific suppression of RMP might be expected if carbamazepine was augmenting the persis-
repetitive firing of CA1 pyramidal neurons along the longitudinal hip- tent sodium current. This suggests that the effect on RMP is either
pocampal axis in response to 100 μM CBZ. In the presence of carba- mechanistically unrelated to the decreased firing frequency in DHC
mazepine, the F–I relationships for DHC neurons were significantly neurons, or that our inferred mechanism involving the persistent so-
more suppressed than their VHC counterparts in both the one and 10 s dium current is incorrect, and that carbamazepine’s dual effects on DHC
time frames (Figs. 2 and 3), despite having similar baseline intrinsic neurons are coupled through an unknown mechanism. We favor the
properties (see above). Moreover, single action potentials were gen- former explanation, as DHC neurons are known to express secondary
erally unaffected by the presence of 100 μM CBZ, although there was a targets for carbamazepine: A1 adenosine receptors (Lee et al., 1983;
slight reduction in the maximal dV/dt for both DHC and VHC neurons Kim and Johnston, 2015). Carbamazepine interacts with A1 adenosine
(Fig. 5). Instead, carbamazepine influenced action potential trains by receptors, which, through their coupling with GIRK channels, hy-
augmenting spike frequency adaptation in DHC neurons (Figs. 2, 3 and perpolarize the membrane potential of DHC neurons (Weir et al., 1984;
6). This enhancement of spike frequency adaptation was so profound, Kim and Johnston, 2015). The slight depolarization of RMP observed in
that most DHC neurons fired only 1–5 action potentials shortly after the DHC neurons could plausibly stem from carbamazepine’s interaction
stimulus onset, which precluded the determination of the SFA ratio for with this system, although additional work will be required to fully
these neurons. This difference between carbamazepine’s effect on DHC explore this hypothesis.
and VHC neurons is striking, and raises questions about mechanistic
differences between these two populations of neurons. Recently, Malik 4.4. Physiological implications
et al. (2016) reported a gradient in spike frequency adaptation along
the longitudinal hippocampal axis, with DHC neurons showing the Although carbamazepine is most commonly used for seizure control
greatest degree of adaptation (Malik et al., 2016). Our work corrobo- in epileptic patients, it is often used to treat non-epileptic conditions,
rates this baseline difference in spike frequency adaptation, and sug- such as neuropathic pain (trigeminal neuralgia and diabetic neuro-
gests that carbamazepine exaggerates this pre-existing difference in pathy), bipolar disorder, and post-traumatic stress disorder (PTSD),
spike frequency adaptation across the longitudinal hippocampal axis, among others (Spina and Perugi, 2004). Because this study was per-
thereby implying some difference in the molecular targets present in formed using naive, non-epileptic tissue, the physiological ramifications
these neurons. of this work should be considered in the context of carbamazepine’s
Carbamazepine’s classical mechanism of action involves the use- uses in non-epileptic patients. Carbamazepine is generally associated
dependent blockade of voltage-gated sodium channels. This progres- with several negative side effects, including significant changes in
sively limits the fraction of available sodium channels throughout an cognitive function, such as reductions in information processing speed,
action potential train, thereby depolarizing threshold and prolonging attention, verbal fluency, and arithmetic (Eddy et al., 2011). Carba-
the interspike interval, which results in spike frequency adaptation. mazepine’s effect on memory is less clear, as some reports claim, im-
This effect relies primarily on the loss of the transient component of pairment, no effect, or even an enhancement in memory tasks (Eddy
sodium currents mediated by voltage-gated sodium channels, due to a et al., 2011). In adult rats similar to those used in this study, carba-
hyperpolarizing shift in the steady-state inactivation curve and corre- mazepine has been shown to impair performance in the Morris Water
spondingly slowed recovery from inactivation. However, Uebachs et al. Maze–a spatial learning and memory task known to rely on the DHC
(2010) have reported a paradoxical carbamazepine-induced effect on (Churchill et al., 2003). This observation is generally consistent with
the persistent sodium current that relies on sodium channel subunit our main finding that carbamazepine’s effects are concentrated within
composition (Uebachs et al., 2010). Specifically, carbamazepine in- the DHC, and suggests that carbamazepine’s influence over the DHC
duces a hyperpolarizing shift in the conductance vs voltage relationship could impact behavior by impairing spatial learning in adult rats.
(G–V curve) of the persistent sodium current in CA1 pyramidal neurons Finally, it is surprising that the carbamazepine-induced suppression
from mice lacking the β1 auxiliary subunit. This augmentation of the of repetitive firing is greater in the DHC, rather than in the VHC, where
subthreshold sodium current accelerates the depolarizing envelope seizures tend to originate, and where CA1 neurons show greater in-
between action potentials, thereby decreasing the interspike interval trinsic excitability. This observation generally resembles another recent
and counteracting the suppression of repetitive firing in these neurons. study by Hönigsperger et al. (2015), which demonstrated a DHC-spe-
The resulting F–I relationships for Scn1b knock-out mice are remarkably cific effect of retigabine–another anti-epileptic drug with a mechanism
unchanged by exposure to 100 μM CBZ (Uebachs et al., 2010). Our unrelated to carbamazepine (i.e., enhancement of IM). Together, these
results are generally consistent with such a mechanism, as the sig- observations suggest that the suppression of activity in the dorsal hip-
nificantly reduced carbamazepine sensitivity of VHC neurons mirrored pocampus could play a larger role in limiting seizure generation than
the effect of 100 μM CBZ on CA1 neurons from Scn1b knock-out mice, previously thought. However, additional experiments exploring the
whereas carbamazepine’s suppression of repetitive firing in DHC neu- effect of anti-epileptic drugs in the dorsal hippocampus of epileptic
rons resembled the Scn1b wild type mouse (Uebachs et al., 2010). Im- tissue will be required to fully explore this issue.
portantly, this potential mechanism is undergirded by the observation
that β1, β3, and β4 sodium channel subunits were expressed more 4.5. Conclusions
abundantly in the DHC relative to the VHC (Table 1), with no difference
in β2-subunit expression or any of the relevant sodium channel α-sub- In summary, we have demonstrated that acute exposure to the use-
unit genes (i.e., Scn1A/Nav1.1, Scn2a1/Nav1.2, Scn3 A/Nav1.3, and dependent sodium channel blocker carbamazepine induces greater
Scn8 A/Nav1.6; Table 1). This observation implies a gradient in β1- suppression of repetitive firing in CA1 neurons from the DHC than the
subunit expression along the longitudinal hippocampal axis, which VHC. Although single action potentials are generally unaffected, car-
could modify carbamazepine’s efficacy along this axis, in a manner bamazepine induces profound spike frequency adaptation in DHC
consistent with our results. The potential influence of the β3 and β4- neurons. Moreover, carbamazepine induces a small, but significant
subunits on carbamazepine efficacy in the DHC, however, remains depolarization of the RMP for DHC neurons, but not VHC neurons, and
unexplored. the magnitude of this depolarization correlates with the suppression of
In addition to the suppression of the F–I relationship, we also ob- repetitive firing across all regions of the longitudinal hippocampal axis
served a slight depolarization of the RMP in DHC neurons, which cor- under investigation. Together, these observations provide novel insight
related with the decreased firing frequency across all cells (DHC and into carbamazepine’s side effect profile in non-epileptic tissue, and
71
M.C. Evans, K.A. Dougherty Epilepsy Research 145 (2018) 63–72
could inform future work on carbamazepine’s (or related anti-epileptic Natl. Acad. Sci. U. S. A. 106, 11794–11799.
drugs) ability to suppress seizure generation in epileptic hippocampal Dougherty, K.A., Islam, T., Johnston, D., 2012. Intrinsic excitability of CA1 pyramidal
neurones from the rat dorsal and ventral hippocampus. J. Physiol. 590, 5707–5722.
tissue. Dougherty, K.A., Nicholson, D.A., Diaz, L., Buss, E.W., Neuman, K.M., Chetkovich, D.M.,
Johnston, D., 2013. Differential expression of HCN subunits alters voltage-dependent
Funding gating of h-channels in CA1 pyramidal neurons from dorsal and ventral hippocampus.
J. Neurophysiol. 109, 1940–1953.
Eddy, C.M., Rickards, H.E., Cavanna, A.E., 2011. The cognitive impact of antiepileptic
This research was supported through funding from Rhodes College. drugs. Ther. Adv. Neurol. Disord. 4, 385–407.
Engel, J., Henry, T.R., Risinger, M.W., Mazziotta, J.C., Sutherling, W.W., Levesque, M.F.,
Phelps, M.E., 1990. Presurgical evaluation for partial epilepsy: relative contributions
Conflicts of interest of chronic depth-electrode recordings versus FDG-PET and scalp-sphenoidal ictal
EEG. Neurology 40, 1670–1677.
None. Fanselow, M.S., Dong, H.-W., 2010. Are the dorsal and ventral hippocampus functionally
distinct structures? Neuron 65, 7–19.
Hönigsperger, C., Marosi, M., Murphy, R., Storm, J.F., 2015. Dorsoventral differences in
Acknowledgements Kv7/M-current and its impact on resonance, temporal summation and excitability in
rat hippocampal pyramidal cells. J. Physiol. 593, 1551–1580.
We would like to thank Dr Darrin Brager and members of the Kim, C.S., Johnston, D., 2015. A1 adenosine receptor-mediated GIRK channels contribute
to the resting conductance of CA1 neurons in the dorsal hippocampus. J.
Dougherty lab for their continuous support throughout this project. Neurophysiol. 113, 2511–2523.
Kwan, P., Sander, J.W., 2004. The natural history of epilepsy: an epidemiological view. J.
Appendix A. Supplementary data Neurol. Neurosurg. Psychiatry 75, 1376–1381.
Lee, K.S., Reddington, M., Schubert, P., Kreutzberg, G., 1983. Regulation of the strength
of adenosine modulation in the hippocampus by a differential distribution of the
Supplementary material related to this article can be found, in the density of A1 receptors. Brain Res. 260, 156–159.
online version, at doi:https://doi.org/10.1016/j.eplepsyres.2018.05. Malik, R., Johnston, D., 2017. Dendritic GIRK channels gate the integration window,
plateau potentials, and induction of synaptic plasticity in dorsal but not ventral CA1
014. neurons. J. Neurosci. 37, 3940–3955.
Malik, R., Dougherty, K.A., Parikh, K., Byrne, C., Johnston, D., 2016. Mapping the elec-
References trophysiological and morphological properties of CA1 pyramidal neurons along the
longitudinal hippocampal axis. Hippocampus 26, 341–361.
Marcelin, B., Liu, Z., Chen, Y., Lewis, A.S., Becker, A., McClelland, S., Chetkovich, D.M.,
Babb, T.L., Brown, W.J., Pretorius, J., Davenport, C., Lieb, J.P., Crandall, P.H., 1984. Migliore, M., Baram, T.Z., et al., 2012. Dorsoventral differences in intrinsic properties
Temporal lobe volumetric cell densities in temporal lobe epilepsy. Epilepsia 25, in developing CA1 pyramidal cells. J. Neurosci. 32, 3736–3747.
729–740. Milior, G., Di Castro, M.A., Pepe’Sciarria, L., Garofalo, S., Branchi, I., Ragozzino, D.,
Bialer, M., Twyman, R.E., White, H.S., 2004. Correlation analysis between anticonvulsant Limatola, C., Maggi, L., 2016. Electrophysiological properties of CA1 pyramidal
ED50 values of antiepileptic drugs in mice and rats and their therapeutic doses and neurons along the longitudinal axis of the mouse hippocampus. Sci. Rep. 6, 38242.
plasma levels. Epilepsy Behav. 5, 866–872. Ragsdale, D.S., Scheuer, T., Catterall, W.A., 1991. Frequency and voltage-dependent in-
Cembrowski, M.S., Bachman, J.L., Wang, L., Sugino, K., Shields, B.C., Spruston, N., hibition of type IIA Na++ channels, expressed in a mammalian cell line, by local
2016a. Spatial gene-expression gradients underlie prominent heterogeneity of CA1 anesthetic, antiarrhythmic, and anticonvulsant drugs. Mol. Pharmacol. 40, 756–765.
pyramidal neurons. Neuron 89, 351–368. Spina, E., Perugi, G., 2004. Antiepileptic drugs: indications other than epilepsy. Epileptic
Cembrowski, M.S., Wang, L., Sugino, K., Shields, B.C., Spruston, N., 2016b. HippoSeq: a Disord. 6, 57–75.
comprehensive RNA-Seq database of gene expression in hippocampal principal Strange, B.A., Witter, M.P., Lein, E.S., Moser, E.I., 2014. Functional organization of the
neurons. eLife 5, e14997. hippocampal longitudinal axis. Nat. Rev. Neurosci. 15, 655–669.
Churchill, J.D., Fang, P.-C., Voss, S.E., Besheer, J., Herron, A.L., Garraghty, P.E., 2003. Uebachs, M., Opitz, T., Royeck, M., Dickhof, G., Horstmann, M.-T., Isom, L.L., Beck, H.,
Some antiepileptic compounds impair learning by rats in a morris water maze. Integr. 2010. Efficacy loss of the anticonvulsant carbamazepine in mice lacking sodium
Physiol. Behav. Sci. 38, 91–103. channel beta subunits via paradoxical effects on persistent sodium currents. J.
Clemens, A.M., Johnston, D., 2014. Age- and location-dependent differences in store Neurosci. 30, 8489–8501.
depletion-induced h-channel plasticity in hippocampal pyramidal neurons. J. Weir, R.L., Padgett, W., Daly, J.W., Anderson, S.M., 1984. Interaction of anticonvulsant
Neurophysiol. 111, 1369–1382. drugs with adenosine receptors in the central nervous system. Epilepsia 25, 492–498.
Dong, H.-W., Swanson, L.W., Chen, L., Fanselow, M.S., Toga, A.W., 2009. Genomic-
anatomic evidence for distinct functional domains in hippocampal field CA1. Proc.
72