Retracted: Cav2.3 (R-type) calcium channels are critical for mediating anticonvulsive and neuroprotective properties of lamotrigine in vivo



This article is corrected by:

  1. Errata: Retraction Volume 57, Issue 11, 1947, Article first published online: 21 March 2016



Lamotrigine (LTG) is a popular modern antiepileptic drug (AED); however, its mechanism of action has yet to be fully understood, as it is known to modulate many members of several ion channel families. In heterologous systems, LTG inhibits Cav2.3 (R-type) calcium currents, which contribute to kainic-acid (KA)–induced epilepsy in vivo. To gain insight into the role of R-type currents in LTG drug action in vivo, we compared the effects of LTG to two other AEDs in Cav2.3-deficient mice and controls on KA-induced seizures.


Behavioral seizure rating and quantitative electrocorticography were performed after injection of 20 mg/kg (and 30 mg/kg) KA. One hour before KA injection, mice were pretreated with 30 mg/kg LTG, 50 mg/kg topiramate (TPM), or 30 mg/kg lacosamide (LSM).

Key Findings

Ablation of Cav2.3 reduced total seizure scores by 28.6% (p = 0.0012), and pretreatment with LTG reduced seizure activity of control mice by 23.2% (p = 0.02). In Cav2.3-deficient mice, LTG pretreatment increased seizure activity by 22.1% (p = 0.018) and increased the percentage of degenerated CA1 pyramidal neurons (p = 0.02). All three AEDs reduced seizure activity in control mice; however, only the non–calcium channel modulating AED, LSM, had an anticonvulsive effect in Cav2.3-deficient mice. Furthermore, LTG altered electrocorticographic parameters differently in the two genotypes: decreasing relative power of ictal spikes in control mice but increasing relative power of high frequency fast ripple discharges during seizures in Cav2.3-deficient mice.


These findings provided the first in vivo evidence for an essential role for Cav2.3 in LTG pharmacology and shed light on a paradoxical effect of LTG in their absence. Furthermore, LTG appears to promote ictal activity in Cav2.3-deficient mice by increasing high frequency components of seizures, resulting in increased neurotoxicity in the CA1. This paradoxical mechanism, possibly reflecting rebound hyperexcitation of pyramidal CA1 neurons after increased inhibition, may be key in understanding LTG-induced seizure aggravation observed in clinical practice.

Today lamotrigine (LTG) is among the most prescribed antiepileptic drugs (AEDs) worldwide. In addition, LTG is approved by the U.S. Food and Drug Administration(FDA) for treatment of bipolar disorder and has become a popular off-label drug for treatment of other neurologic and psychiatric conditions such as borderline personality disorder. This diverse therapeutic capacity of LTG probably reflects the nonspecificity of the drug, which is known to inhibit several different calcium, potassium, and sodium currents (Beck & Yaari, 2012). LTG is thought to mediate its anticonvulsant and neuroprotective effects in vivo predominantly by inhibiting voltage-dependant sodium currents and the subsequent glutamate release; however, recent evidence suggests that in mice, inhibition of Cav2.3 channels could play an important role in the mechanism of action of LTG during experimentally induced epilepsy. It has been demonstrated that LTG and another modern AED, topiramate (TPM), inhibit R-type currents in heterologous systems and brain slices (Hainsworth et al., 2003; Kuzmiski et al., 2005). Furthermore, Cav2.3-deficient (Cav2.3-KO) mice display seizure resistance and reduced hippocampal neurotoxicity after kainic acid (KA) injection (Weiergräber et al., 2007). Parenteral administration of KA is a well-established method of modeling temporal lobe epilepsy, causing seizures in the hippocampus and temporal lobe and degeneration of hippocampal pyramidal neurons (Tremblay & Ben Ari, 1984; Sperk et al., 1985). Using the KA model of temporal lobe epilepsy, we investigated the effect of LTG next to TPM and lacosamide (LSM) in Cav2.3-KO and controls. The new AED lacosamide (LSM), which enhances slow inactivation of voltage-gated sodium channels(Errington et al., 2008), was used as a positive control, as it has been shown not to effect calcium or potassium currents (Errington et al., 2006; Wang & Khanna, 2011). Gaining insight into the role of Cav2.3 calcium channels in antiepileptic pharmacotherapy may allow identification of new antiepileptic mechanisms and therefore of novel potential drug targets, offering hope for patients with drug refractory epilepsy.

Materials and Methods


Cav2.3-KO and control mice are separate mouse lines derived from heterozygous parents (fourth backcrossing into C57Bl/6). Homozygous littermates are regularly interbred with each other and back-bred into C57Bl/6 (for further information on knock out generation see (Pereverzev et al., 2002; Weiergräber et al., 2006). Mice were kept at 20°C in polycarbonate cages under a 12 h light–dark cycle (7:00 a.m./p.m.) with food and water ad libitum. All animal experiments were in line with the European Communities Council Directive for the care and use of laboratory animals and were approved by the local institutional committee on animal care.

Antiepileptic pretreatment and seizure induction

Between 9:00 and 10:30 a.m., saline or AEDs LTG (30 mg/kg), TPM (50 mg/kg) (both Sigma Aldrich, Crailsheim, Germany), or LSM (30 mg/kg) (UCB Pharma SA, Brussels, Belgium) dissolved in saline were injected intraperitoneally into male mice of both genotypes (Cav2.3-KO and Cav2.3+|+) within the age range of 20–25 weeks. One hour later 20 mg/kg KA (n = 40) or 30 mg/kg KA (n = 46) (Sigma Aldrich) dissolved in saline was injected intraperitoneally. Immediately after administering KA, mice were set in separate single cages and filmed for the next 24 h.

Behavioral seizure analysis

The initial 2 h after the application of KA was evaluated using an adapted version of Morrison's Seizure Rating Scale (Morrison et al., 1996) (see Data S1 for further information):

  • Stage 0: normal behavior.
  • Stage 1: immobility.
  • Stage 2: facial clonus, head bobbing/nodding, automatisms.
  • Stage 3: limb clonus, jerking of the torso.
  • Stage 4: rearing.
  • Stage 5: falling.
  • Stage 6: tonic–clonic seizure.
  • Stage 7: tonic–clonic seizure with jumping.
  • Stage 8: tonic–clonic seizure causing death.

For a 2 h period the highest seizure score was noted for every 5-min interval. Interval scores were added for total seizure scores. Total seizure scores of pretreated groups were compared to those of the untreated group of the same genotype and differences were expressed as relative changes in total seizure score from the untreated group.

Radiotelemetric electrocorticographic recording of seizures

Radiotelemetric electrocorticography (ECoG) of KA-induced seizures was recorded on LTG pretreated and untreated animals of both genotypes (n = 4 per group). Animals were anesthetized with 100 mg/kg body weight (BW) ketamine hydrochloride (Ketanest, Parke-Davis/Pfizer, Berlin, Germany) and 10 mg/kg BW xylazine hydrochloride (RompunR 2%; Bayer Vital, Leverkusen, Germany). TL11M2-F20-EET transmitters (Datascience International, Lexington, MA, U.S.A.) were implanted subcutaneously and burr holes were drilled over the somatosensory cortex (−1 mm and 3 mm lateral from bregma) and cerebellum (−6.3 mm and 1 mm lateral from bregma), leaving the dura intact. Electrodes were inserted and fixed into position with glass ionomer cement (Kent DentalR, Kent Express, Kent, United Kingdom). Animals were allowed 7 days to recover from surgery (all made full recovery) and were then recorded before (control condition) and after injection of 20 mg/kg KA, i.p. ECoG studies were obtained at a sampling rate of 1,000 Hz without cutoff from freely moving animals in their cages, which were placed on the telemetry receiver platforms.

ECoG analysis

NEUROSCORE 2.1.0 (Datascience International) was used to calculate absolute and relative power of frequency bands (Fast Fourier Transform based using a Hamming window) in the first hour after KA injection (totally and fractioned into 5-min intervals). The frequency spectrum was defined as follows: Delta (0.5–4 Hz), Theta (4–8 Hz), Alpha (8–12 Hz), Sigma (12–16 Hz), Beta (16–24 Hz), Gamma (30–80 Hz), Ripples (80–200 Hz), and Fast Ripples (200–500 Hz). An automated seizure detection protocol was written to quantify ictal activity. The protocol recognizes waveforms shorter than 200 msec that are between 2.5- and 25-fold the baseline amplitude as spikes. Spikes occurring in intervals between 30 and 1,500 msec are recognized as belonging to a spike train, which must be at least 300 msec long and contain a minimum of four spikes. No ictal events were detected in the control condition (before KA injection). The Z-ratio reflecting the ratio between low and high frequency power (LF [0.5–8 Hz] and HF [8–20 Hz], respectively) was calculated using the following equation: (LF − HF)/(LF + HF).

Histology and immunohistochemistry

Seven days after injection of 30 mg/kg, KA brains were extracted and kept in 30% sucrose for 24 h prior to freezing them in methyl-butane. Brains were sliced 10 μm thick in a cryotome (CM3050S; Leica Microsystems, Wetzlar, Germany), and then fixed in 4% formaldehyde and Nissl stained according to standard protocol. Brains for immunohistochemistry were kept in 4% paraformaldehyde for 24 h and—using a slow regimen of manual changes over 12 days—embedded in paraffin. Ten micron sections were cut with slim Feather blades for low compression (cutting angle 25 degrees) on a motor-driven rotary microtome (Reichert-Jung 1140 Autocut, Leica Microsystems, Nussloch, Germany) and mounted on silanized glass slides. Sections were deparaffinized and rehydrated before incubation with anti-neuron specific nuclear protein (anti-NeuN) antibody from mouse (GeneTex Asia Ltd, Hsinchu City, Taiwan) and detection thereof using VECTOR M.O.M Peroxidase Immunodetection Kit (Vector Laboratories Inc, Burlingame, CA, U.S.A.). Using the cell counter tool of NIH IMAGEJ software (,hippocampal neurons were counted and the percentage of pyknotic neurons calculated.

Protein isolation, Western blot analysis, and protein quantification

Twenty-four hours after 30 mg/kg KA (or saline) injection, membrane proteins were isolated from control mouse (n = 10) hippocampi using a high-salt high-pH extraction method (for further information see [Wisniewski, 2009]). Fifty micrograms of membrane protein per sample were separated by electrophoresis on an sodium dodecyl sulfate polyacrylamide gel and then blotted onto a polyvinylidene fluoride membrane. The Cav2.3 calcium channel was detected using a self-generated antibody (rabbit) directed against AA 256–272 in the loop IS5 to pore region of the human alpha1E subunit (for further information see [Pereverzev et al., 1998]), ECL-Anti-Rabbit IgG and ECL detection system (GE Healthcare, Buckinghamshire, United Kingdom). Because the expression of the reference protein synaptophysin (SYN) has been shown to be unaffected by hyperexcitation (Chen et al., 2001a; Wierschke et al., 2010), Cav2.3 bands were quantified by normalizing them to SYN, which was detected using anti-SYN antibody from mouse (Antibodies-online, Atlanta, GA, U.S.A.) and ECL-Anti-Mouse IgG (GE Healthcare). Cav2.3 protein was quantified manually using IMAGEJ 1.46 (NIH) and automatically using GELSCAN 6.0 (BioSciTec, Frankfurt, Germany).

Statistical analysis

Seizure scores and relative spectral power were assessed using the Shapiro-Wilk test of normality and found to be mostly nonnormally distributed. Therefore, the nonparametric Mann-Whitney test was used to determine significance of seizure scores. Relative power values were log transformed (log(x/[1 − x])) to obtain a more Gaussian distribution and were then subjected to analysis of variance (ANOVA) (Gasser et al., 1982). Statistical significance of frequencies of the seizure stages was determined using Fisher's exact probability test. p-Values of 0.05 and below were considered statistically significant.


Behavioral seizure analysis

After injection of 20 mg/kg KA in all groups, normal explorative behavior ceased within 10 min and mice “froze” exhibiting a rigid posture and staring into space (immobility stage i.e., stage 1). In this stage, mice only reacted scarcely to their environment (i.e., when nudged) if at all. Six of eight control mice experienced tonic–clonic seizures (Fig. 1A), whereas Cav2.3-KO mice did not develop tonic–clonic seizures or enter seizure stages higher than stage 3 (Fig. 1E), displaying a reduction of total seizure scores of 28.6% compared to control mice (from 57.8 ± 2.6 to 41.4 ± 3.7, p = 0.0012; U = 2.5) (Fig. 2A). In control animals, LTG prevented tonic–clonic seizures (Fig. 1B) and reduced total seizure scores by 23.2%, from 57.8 ± 2.6 to 44.3 ± 3.6 (p = 0.02; U = 6.5). TPM did not prevent tonic–clonic seizures in all control mice (Fig. 1C) but reduced total seizure scores by 21%, from 57.8 ± 2.6 to 45.6 ± 3.8 (p = 0.029; U = 5). LSM was most effective in reducing seizure scores in control mice, eliciting a reduction of the total seizure score of 42.2%, from 57.8 ± 2.6 to 33.4 ± 2.5 (p = 0.0016; U = 0) (Fig. 1D). TPM had no significant effect on total seizure scores in Cav2.3-KO mice, whereas LTG significantly increased total seizure scores by 22.1%, from 41.4 ± 3.7 to 50.6 ± 1.5 (p = 0.018; U = 6.5) and the frequency of the convulsive stage 3 (Fig. 1F and Table S1) in Cav2.3-KO mice. Both LTG and TPM were effective in reducing total seizure scores of control mice but were ineffective in doing the same in Cav2.3-KO mice. LSM was the only AED of the three that reduced seizure scores in Cav2.3-KO mice, doing so by 19.4%, from 41.4 ± 3.7 to 33.4 ± 0.6 (p = 0.048; U = 5) (Fig. 2A). No animals died as a result of 20 mg/kg kainic acid injection. In control mice, all three AEDs significantly increased the frequency of stage 1, the lowest pathologic seizure stage, whereas LTG had the opposite effect in Cav2.3-KO mice (Table S1A). TPM did not alter the frequencies of occurrence of the seizure stages in Cav2.3-KO mice.

Figure 1.

Individual seizure scores of mice at 20 mg/kg kainic acid. Seizure scores are plotted for each mouse during the course of the 2 h video-monitoring period. (AD) Seizure scores for control mice without (A) and after pretreatment with LTG (B), TPM (C), or LSM (D). (EH) Seizure scores for Cav2.3KO without (E) and after pretreatment with LTG (F), TPM (G), or LSM (H).

Figure 2.

Effect of LTG, TPM, and LSM on experimentally induced epilepsy in Cav2.3-deficient and control mice. Seizures were induced in n = 7 Cav2.3KO and n = 8 control mice by intraperitoneal administration of KA. The animals were video-monitored for 2 h after injection. A semiquantitative scale was used as described in methods. Cav2.3KO and control mice were injected with 30 mg/kg LTG (n = 7 and n = 8, respectively) or 50 mg/kg TPM (n = 6 and n = 6, respectively), or 30 mg/kg LSM (n = 5 and n = 5, respectively) 1 h before KA injection. (A) Total seizure scores in control and Cav2.3-KO mice with and without antiepileptic pretreatment after 20 mg/kg KA. All AEDs reduce total seizure scores in control mice; however, only LSM reduces total seizure scores in Cav2.3KO mice. Note, Cav2.3KO mice reach significantly higher scores under LTG than without pretreatment. (B) Effects of LTG on total seizure scores after 30 mg/kg KA injection are comparable to those after 20 mg/kg.

We retested the effect of LTG in Cav2.3KO and control mice at 30 mg/kg KA (Fig. 2B), a dosage at which Cav2.3KO mice develop tonic–clonic seizures and exhibit similar seizure activity as control animals at 20 mg/kg KA, to determine whether LTG can prevent tonic–clonic seizures in Cav2.3KO mice and to further investigate the convulsive effect of LTG in Cav2.3KO mice observed at 20 mg/kg. At 30 mg/kg KA, LTG pretreatment reduced total seizure scores of control mice by 30% (p = 0.0079; U = 0) and total seizure scores of LTG-treated Cav2.3-KO mice were 33% (p = 0.015; U = 1.5), higher than those of LTG-treated control mice (69 ± 6.4 compared to 51.6 ± 1). An increase (not significant) of total seizure scores of 15.8% (69 ± 6.4 compared to 59.6 ± 4.1) was observed in LTG-treated Cav2.3-KO mice compared to Cav2.3-KO mice without pretreatment, which is in line with the significant increase of total seizure scores caused by LTG in Cav2.3-KO mice observed at 20 mg/kg KA. At both KA concentrations, LTG increased the frequency of stage 3 (Tables S1A,B) in Cav2.3KO mice (but not in controls), which contributes to the increased total seizure scores of LTG-treated Cav2.3-KO mice compared to untreated Cav2.3-KO mice.

Histology and immunohistochemistry

NeuN- and Nissl-stained brain sections of mice from 30 mg/kg groups (n = 4 per group) were evaluated by determining the percentage of pyknotic to healthy pyramidal neurons in the CA1 (Fig. 3) CA2, CA3, and dentate gyrus (DG) regions of the hippocampus. Cav2.3-KO mice were found to display significantly less pyknotic pyramidal neurons than control mice in the CA1 and CA3 regions of the hippocampus (CA1 4.14 ± 2.07% compared to 26.5 ± 6.41%; CA3 6.89 ± 0.75% compared to 27.17 ±4.75%), which is in line with findings from Weiergräber et al. (2007). Both stains revealed that LTG-treated Cav2.3-KO mice displayed significantly increased degeneration of pyramidal CA1 neurons compared to untreated Cav2.3KO mice (NeuN 14.65 ± 3.45% compared to 4.14 ± 2.07%, p = 0.048; Nissl 20.6 ± 2.6% compared to 11 ± 2%, p = 0.02;), although a similar trend is visible in the other three regions. Furthermore, in control mice, LTG significantly reduced neurodegeneration in the CA1, CA3, and DG. Cav2.3KO and LTG-treated control mice displayed similar degrees of degeneration in all evaluated regions except the CA2.

Figure 3.

Neurotoxicity after 30 mg/kg KA. NeuN-stained paraffin sections of the CA1 (A), CA2 (B), CA3 (C), and DG (D) regions of the hippocampus at 20× magnification from control mice (top left of each quadrant), LTG pretreated control mice (bottom left), Cav2.3KO mice (top right), and LTG pretreated Cav2.3KO mice (bottom right). Higher (40×) magnification of NeuN- (E) and Nissl- (F) stained sections showing the CA1 region. Arrows indicate pyknotic pyramidal neurons, that is, pyramidal neurons with condensed nuclear material indicating apoptosis. (G) Evaluation of the percentage of pyknotic neurons in regions of the hippocampus.

Expression of Cav2.3 protein

Both manual and automated quantification of western blotted Cav2.3 bands by normalization to SYN revealed no significant differences in Cav2.3 protein expression between KA- and saline-injected groups (Fig. 4).

Figure 4.

Cav2.3 protein expression in the hippocampus. (A) Western blot of control mice hippocampal membrane fractions. (B) Student's t-test of manual (using IMAGEJ 1.46) and automated (using GELSCAN 6.0) quantification of Cav2.3 protein by normalization to synaptophysin expression revealed no significant difference between KA- and saline-injected animals. Although manual quantification produced a greater difference between the groups, standard error was also much increased.

Electrocorticography studies

Relative power was used in the evaluation and statistical testing due to better inter-individual comparability; however, absolute power was also computed and is shown in Fig. 5.

Figure 5.

Evolution of absolute power after KA injection. Absolute power of the frequency bands for 5-min epochs after KA injection of control mice (A) with LTG pretreatment (B) and Cav2.3KO mice (C) with LTG pretreatment (D). The robust increase of absolute delta and theta power over time in control mice and LTG-pretreated Cav2.3KO mice represents the genesis of ictal discharges, which occur predominantly in these two frequency bands. Note the effect of LTG on HF bands in both genotypes.

Effect of LTG in control condition

Spectral analysis of the recorded ECoG studies revealed significant differences between Cav2.3-KO and control mice and between the effects of LTG in both genotypes in control recordings and after injection of 20 mg/kg KA. In control conditions, Cav2.3-KO mice displayed significantly reduced relative delta power compared to control mice (29 ± 1.7% vs. 22.1 ± 2.0% [p = 0.037]) (Fig. S1A). LTG treatment increased relative beta power in control mice (from 4.6 ± 0.48% to 7.5 ± 1% [p = 0.037]) (Fig. S1B), but not in Cav2.3-KO mice in which LTG reduced relative alpha power from 16 ± 1.5% to 11.1 ± 0.9% (p = 0.034) (Fig. S1C).

Effect of KA compared to control condition

KA injection elicited spikes, sharp waves, and spike trains in all four groups, with ictal activity predominantly occurring within the delta–theta range (Fig. 6). Accordingly, KA injection significantly increased relative delta power in both genotypes (Fig. S2A,C), however, to a greater degree in control mice. Of interest, in control mice, LTG pretreatment prevented the KA-induced shift in spectral distribution (Fig. S2B), whereas in LTG-pretreated Cav2.3-KO mice, KA injection caused a significant reduction of alpha power (Fig. S2D).

Figure 6.

Ictal activity in ECoG recordings after 20 mg/kg KA. Raw ECoG traces of seizures in individual mice of each group. Blue dots indicate individual ictal spikes; green lines indicate spike trains.

Effect of LTG on KA-induced seizures

Both genotypes displayed different spectral distribution after KA injection (Fig. 7), with Cav2.3-KO mice exhibiting significantly increased relative sigma and beta power compared to control mice (4 ± 0.2% vs. 3.2 ± 0.1% [p = 0.009] and 4.9 ± 0.4% vs. 3.5 ± 0.3% [p = 0.024], respectively), reflecting less ictal activity in the delta theta range and thus the reduced seizure susceptibility found by other authors (Weiergräber et al., 2007) and observed in behavioral analysis in this study.

Figure 7.

Spectral distribution after KA injection. Relative power of the frequency bands after KA injection averaged from 10-s epochs of 60-min recording time. (A) Cav2.3KO versus control mice. (B) Control versus LTG-pretreated control mice. (C) Cav2.3KO versus LTG pretreated Cav2.3KO mice. Note the similar effect of ablation of Cav2.3 and LTG pretreatment (of control mice).

Similarly LTG-pretreated control mice exhibited reduced relative theta power compared to untreated control mice (31 ± 1.9% vs. 20.4 ± 2.7% [p = 0.04]), and therefore a distinct shift in spectral distribution toward sigma and beta frequencies (5.4 ± 0.9% vs. 3.2 ± 0.1% [p = 0.007] and 6.5 ± 1% versus 3.5 ± 0.3% [p = 0.008], respectively) away from delta–theta frequencies and thus less ictal activity in this frequency range. In contrast, LTG pretreatment of Cav2.3-KO mice did not significantly alter spectral distribution when the complete recording period was analyzed.

Analysis of maximal seizure activity

Using the automated spike detection protocol, the longest spike train, that is, maximal seizure activity, was identified and analyzed in further detail in order to gain more detailed insight into the effect of LTG on ictal activity in both genotypes. Although analysis of the parameters latency to first spike, longest spike train, spikes per second, and average spike interval revealed trends corresponding to the rest of the data, results did not reach statistical significance, as inter-individual spiking patterns proved to be highly variable within the groups. However, analysis of maximal seizure activity revealed a robust reduction of relative delta power in LTG-pretreated control mice compared to those without pretreatment (51 ± 7.2% vs. 24.9 ± 2.6% [p = 0.03]) (Fig. 8B). This effect of LTG on maximal seizure activity did not occur in Cav2.3-KO mice.

Figure 8.

Spectral distribution during maximal seizure activity. Relative power of the frequency bands during the longest spike train identified by the seizure detection protocol averaged from 10-s epochs. (A) Cav2.3KO versus control mice. (B) Control versus LTG-pretreated control mice. (C) Cav2.3KO versus LTG-pretreated Cav2.3KO mice. LTG increases fast ripples and fails to reduce spiking in the delta theta range in Cav2.3KO mice as it does in control mice.

Of interest, LTG-pretreated Cav2.3-KO mice displayed significantly increased relative fast ripple power compared to untreated Cav2.3-KO mice (1.2 ± 0.4% vs. 0.036 ±0.001% [p = 0.003]), possibly underlying the pro-ictogenic effect of LTG observed in behavioral seizure analysis (Fig. 8C). Correspondingly, in control mice but not in Cav2.3-KO mice, LTG significantly reduced the Z-ratio of maximal seizure activity from 0.51 to 0.07 (p = 0.04), indicating an increase of high frequency power and thus a shift away from spiking in the delta–theta range.


In this study we show that the Cav2.3 calcium channel is critical in mediating the anticonvulsant properties of LTG in the KA model of epilepsy and that LTG elicits pro-ictogenic effects in mice lacking the Cav2.3 calcium channel. Neither LTG nor TPM, which have been shown to inhibit R-type currents in heterologous systems, could reduce seizure scores in Cav2.3-KO mice, indicating the importance of Cav2.3 inhibition in mediation of their anticonvulsive effects. In contrast, LSM, which has no calcium channel modulating properties, was the only AED of the three tested that could reduce seizure scores in Cav2.3-KO mice. It should be taken into account that in control mice neither LTG nor TPM was capable of reducing seizure scores beyond the degree that is reached when the Cav2.3 is ablated. Furthermore, this study reveals a convulsive and neurotoxic effect of LTG in the absence of Cav2.3 calcium channels. Of interest, toxicity of LTG was located in the CA1 region of the hippocampus, where LTG is known to be most neuroprotective (Leach et al., 1991; Crumrine et al., 1997; Englund et al., 2011). Therefore, it is assumable that the underlying neuroprotective mechanisms may include inhibition of signaling through Cav2.3, which we found not to be upregulated after KA injection. The fact that the convulsive effect of LTG is more specifically related to the CA1 region, must lead to a novel interpretation of its mechanism of action. Underlying this finding could be postinhibitory rebound firing of CA1 pyramidal neurons promoted by HCN channels (hyperpolarization-activated cyclic nucleotide-gated channels), a paradoxical phenomenon observed as a reaction to increased inhibition after experimentally induced seizures (Chen et al., 2001b). LTG has been shown to enhance HCN currents in CA1 pyramidal neurons, conveying an inhibitory effect (Poolos et al., 2002). However, due to the capacity of HCN channels to activate at hyperpolarized potentials and slow deactivation kinetics, increased synaptic inhibition, a condition predictable in Cav2.3KO mice, may cause rebound excitation of CA1 pyramidal neurons when HCN currents are stimulated by LTG. It should be noted that no compensatory upregulation of other cation channels that may increase excitability was identified after injection of 30 mg/kg KA in hippocampi of Cav2.3-KO mice compared to control mice in a full transcriptome analysis that was performed in our laboratory prior to the present study (results not shown).

Furthermore, in this study, telemetrically recorded ECoG revealed that LTG cannot attenuate ictal discharges in Cav2.3KO mice as it does in control mice, but instead increases ultra-high frequency components of ictal activity, which are known to be associated with generation of epileptic activity in humans and in animals (Allamand et al., 1997; Traub et al., 2001; Bragin et al., 2004). Clinically, this phenomenon observed in mice and in brain slices, may be represented by the capacity of LTG to aggravate seizures in certain epilepsy syndromes. Although toxic doses of several (nonsedative) AEDs can cause seizures, LTG has been reported to cause and aggravate seizures and seizure frequency at doses within its therapeutic range. In severe myoclonic childhood epilepsy, there is a frequent aggravating effect of LTG at therapeutic doses (Guerrini et al., 1998; Genton, 2000). Another study reports that adults with idiopathic generalized epilepsies treated with LTG experienced exacerbation or de novo appearance of myoclonic jerks (Crespel et al., 2005). Whether this paradoxical effect of LTG in clinical practice reflects rebound hyperexcitation after increased inhibition, possibly due to antiepileptic polytherapy or intake of other drugs with an inhibitory effect on certain neuron types, must be investigated in further studies. It is notable that nothing is known about expression or genetic variants of Cav2.3 in human patients with epilepsy. Although gain-of-function mutations in the CACNA1H gene encoding for the low-voltage activated (T-type) calcium channel Cav3.2 have been identified in patients with hereditary forms of absence epilepsy (Liang et al., 2006), no variants of Cav2.3 have been identified to date in patients with epilepsy. However, increased R-type currents have been measured in the genetically epilepsy-prone rat (GEPR), suggesting that increased R-type signaling contributes to the genetic basis of the enhanced seizure susceptibility of GEPR (N'Gouemo et al., 2010). Whether expression of Cav2.3 is altered in the hippocampus of human patients with epilepsy is a matter of great interest; however, gaining access to resected hippocampal tissue can be difficult, and is a limiting factor for several epilepsy researchers. Nevertheless, investigation of genetic variants of CACNA1E in patients with epilepsy who experience a worsening of symptoms with LTG could produce valuable insights.

Because LTG is not able to prevent or attenuate ictal activity in the absence of Cav2.3 calcium channels, one must assume that its anticonvulsive properties are not based primarily on inhibition of sodium currents, but that R-type modulation plays a major role in mediating net anticonvulsive properties of LTG. A complex and multimodal mechanism of LTG is highly likely, also considering that LTG has been shown to attenuate several neuropsychiatric disorders such as bipolar depression, borderline disorder, and anxiety disorder, and to contribute to a better outcome in animal models of stroke and subarachnoid hemorrhage.


This project has been kindly funded by Köln Fortune. We would like to specially thank Mrs. Renate Clemens and Mrs. Nadin Piekarek for their dedication and hard work.


None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.