This study has 3 main findings. 1) Cox-2, TIEG1 and Pcdh-8 are differentially expressed in the hippocampus during the epileptic process in the kindled rat and the differential expression is dependent on the duration of the AD measured by the recording electrode placed in the amygdala. 2) All 3 genes belong to a seizure-induced IEG expression profile that also comprises at least 16 other genes. Many of these genes have previously been described in various models of synaptic plasticity. 3) In our study, we also found that the anti-epileptic drug levetiracetam attenuated the kindling-induced changes of both immediate early and late response gene expression 3 hours after the last seizure. Previously, the effect of levetiracetam on kindling-induced gene expression after 1 hour has been reported for two single genes BDNF and NPY  and after 24 hours for NPY, TRH and GFAP  but never after 3 hours on a larger panel of both IEG and late response genes. Consequently, our data further expands the list of transcripts sensitive to the actions of levetiracetam. In addition, we also found that the attenuating effect of levetiracetam on seizure-induced gene expression was paralleled by the drug effect on AD duration which has not been reported previously. These findings suggest that levetiracetam during epileptic seizures targets mechanisms that reduce the AD duration which in turn prevents the increase in mRNA expression of synaptic plasticity-related genes in areas undergoing synaptic remodelling in response to the enhanced neuronal activity which is reported to occur both in vitro and in vivo [42, 43].
Kindling-induced expression of a subset of genes associated with synaptic remodeling
Utilising a DDRT-PCR based approach to find gene expression changes in the hippocampus during amygdala-kindling Cox-2 was identified as differentially expressed. Cox-2 has previously been associated with epileptic conditions in both animals and humans [44, 45]. Two additional kindling-induced transcripts were identified and validated by the DDRT-PCR method. Of these, TIEG1 was originally identified as differentially expressed in human osteoblasts after TGF-β activation . It is induced in oligodendroglial precursor cells after TGF-β treatment  and in rat brain after kainate-induced seizures . TGF-β is the most potent inducer of TIEG1 but other members of the TGF-β superfamily also induce TIEG1 including GDNF, BMP-2 and β-A activin [49, 50]. In contrast, information on signaling cascades involving the cell-adhesion molecule Pcdh-8 is very limited. Pcdh-8 is predominantly expressed in the nervous system and during development  and it has previously been implicated in neuronal plasticity, long-term potentiation and synaptic remodelling [52, 53].
Subsequently, a bioinformatics approach combined with quantitative RT-PCR lead to the identification of 16 additional kindling-induced genes. The differential expressed transcripts fall into the following functional categories: transcription factors (c-fos, krox-20 and egr3), neurotrophic factors (NGF, BDNF and NT-3), inflammatory response (Cox-2 and TNF-α), extracellular matrix components (Narp and MMP9), cell-adhesion (Pcdh-8), TGF-β superfamily signaling (TIEG1 and β-A activin), GPCR signaling (RGS2, Homer1a and Ania3) and others (Pim1 kinase, Arc and Synaptopodin). Historically, many of these transcripts have been linked to synaptic plasticity and models of enhanced neuronal activity [39, 54–71, 52, 72, 73]. In agreement, DDRT-PCR studies in rapidly kindled mice revealed differential expression of 26 transcripts amongst those RGS2, krox-24, Homer and c-fos  and differential expression of Cox-2, TIEG1, Narp, Arc, Ania-3, Homer1a/Vesl and BDNF are also seen in the hippocampus after electroconvulsive shock . Two of the differentially expressed transcripts Homer-1a and Ania-3 originate from the same Homer 1 genetic loci . However, we rule out a more general transcriptional activation of the Homer 1 gene in amygdala-kindled rats as a third splice variant of the Homer1 gene Homer-1b/c is not differentially expressed after kindling-induced seizures (see Additional file 1, Table S2). Consequently, we propose that the kindling-induced transcripts are molecular indicators of immediate early changes, which are associated with enhanced neuronal activity in the hippocampus in models of synaptic plasticity including epilepsy models. Whether any, just a subset or all of these genes are causal for the kindling process still remains to be verified.
Induction of IEGs in the hippocampus seems to be an "all-or-none" effect
The expression of c-fos mRNA has previously been shown induced ipsilaterally after only a few ADs. However, upon repeated stimulations the kindling-induced c-fos mRNA expression progresses to the contralateral hippocampus in animals at stage 3 or more; at later stages kindled animals exhibit no asymmetry in c-fos expression . Likewise, we also found a good correlation between the ipsi- and contralateral hippocampus in the normalized expression levels of Pcdh-8 and Cox-2 in stage 3 and stage 5 kindled rats. The progression of mRNA expression of these genes from ipsilateral to contralateral regions during progression of kindling at earlier stages than stage 3 also indicates that these genes are induced as a consequence of the spread of seizure activity from ipsilateral to contralateral regions. Overall, we observed no significant difference in the IEG mRNA levels between stage 3, stage 5 and fully kindled populations. However, when analysing individual animals, a higher frequency of IEG mRNA induction was seen at the later stages compared to earlier kindling stages. This could indicate an "all-or-none" effect in transcriptional induction rather than a gradual increase in the transcriptional response when animals progress from stage 3 to fully kindled. At the single animal level kindling-induced IEG expression was observed in half of the stage 5 and 80% of the fully kindled rats, respectively. So even though the incident rate of IEG induction increases with kindling acquisition not all kindled animals have changes in hippocampal IEG expression, which is also in good agreement with previous studies [76, 77]. In addition, at the single animal level there seems not to be a linear correlation between seizure parameters like seizure severity and AD duration and the relative mRNA expression levels of the examined genes (Figure 5) suggesting an "all or none" effect of kindling on IEG induction in the hippocampus. In support, kindled rats with "high IEG" expression also have significant longer AD duration times than kindled rats with "low IEG" expression.
Levetiracetam reduces the probability for induction of IEG transcripts involved in synaptic remodeling in the hippocampus
In line with other studies we show that levetiracetam exhibits an anti-epileptic effect in the amygdala-kindling model, i.e. animals pre-treated with levetiracetam have shorter AD duration and lower seizure severity compared to non-treated controls [11, 12]. In our study, the anti-epileptic effect paralleled the observation that levetiracetam also exerted a potent inhibitory effect on kindling-induced changes in IEG mRNA expression in the hippocampus. However, analysis of the individual animals revealed that only one of the 16 levetiracetam treated animals showed a significant change in the IEG mRNA levels, whereas 9 of the 16 vehicle treated animals exhibited robust changes in IEG mRNA levels. Nevertheless, 5 of 16 levetiracetam treated animals developed stage 5 seizures during the treatment. No difference in the mRNA levels of the levetiracetam target SV2A as well as in a number of other neuronal genes was observed hence ruling out compensatory mechanisms as sensitization/desensitization of SV2A or even a more general toxicity effect of the drug on neuronal expressed mRNAs. Conclusively, this suggests that levetiracetam might exert part of its anti-epileptic effect by reducing the likelihood of an "all-or-none" induction of genes putatively involved in establishing long term synaptic changes important to kindling acquisition, synaptic plasticity and perhaps epileptogenesis.
Expression of the IEG expression is not necessarily induced during neuronal hypersynchronous seizure activity in the hippocampus as it also requires bursts of population spikes . This has been further supported by human studies showing correlation between spiking activity and gene expression changes . Levetiracetam is able to inhibit hypersynchronous activity by reducing the epileptiform activity-induced population spikes in the CA3 area of the hippocampus [79–82]. This suggests that levetiracetam is attenuating the kindling-induced IEG mRNA expression by targeting mechanisms that reduces the epileptiform activity-induced population spikes in the hippocampus. However, it would require a recording electrode in the hippocampus to be able to correlate the AD duration time, existence of population spike activity in the hippocampus and increased IEG expression to the attenuating effects of levetiracetam.