A total of 70 rats underwent stereotaxic surgery. Rats were individually anesthetized with isoflurane (5% initial, 2–2.5% maintenance) and treated an analgesic (carprofen, 10 mg/kg, i.p.) to minimize pain and discomfort during post-surgery recovery. The rat was then secured in a stereotaxic apparatus and an incision was made along the midline of the scalp and the overlying skin retracted. A single bipolar electrode (MS 303/2-B, Plastics One, Roanoke VA) was implanted into the left basolateral amygdala at the following coordinates: −2.8 mm posterior to bregma, +5.00 mm medial/lateral, and −8.5 mm ventral to surface of the skull (Paxinos and Watson, 1998). The electrode assembly was secured with stainless steel screws and dental acrylic. The incision site was cleaned with 10% povidone iodine solution and a topical antibiotic cream (Polysporin^®) was applied to reduce risk of infection. Rats were placed in heated recovery cage until they were fully recovered and then returned to their home cage. Twenty-four hours later, rats received daily treatment with carprofen (10 mg/kg, i.p.) daily for the first 3 to 5 days after surgery to reduce signs of pain and inflammation. Rats were allowed at least 1 week of recovery from surgery before start of kindling.

Rats were randomly assigned to one of three conditions: long-term kindling, long-term kindling with TMZ treatment, or sham controls. In this context, long-term kindling refers to the cumulative or total number of electrical stimulations, which generally exceeds 60 stimulations (Kalynchuk and Fournier, 2009). Unlike conventional kindling procedures, the number of stimulations during long-term or extended kindling exceeds the number required for animals to reach a fully kindled state, which is defined as a minimum of three consecutive Class 5 or higher generalized convulsive seizures. As such, long-term kindling is considered to model more closely the progressive epileptogenic changes and associated comorbidities that are seen in chronic epilepsy (Kalynchuk and Fournier, 2009).

The kindling procedure consisted of 75 or 105 electrical stimulations delivered by an isolated pulse stimulator (Model 2100, A-M Systems, Sequim WA USA) to the left basolateral amygdala (1 msec biphasic square wave pulses, 60 p.p.s for 1 s). The amplitude of the current was set to 800 μA (peak-to-peak). The intensity of this electrical stimulation has been shown to be well above the threshold to evoke an epileptiform afterdischarge in the amygdala (Racine, 1972a). Electrical stimulations were delivered three times daily with a minimum of 3 h between consecutive stimulations. Stimulations were delivered for 4 consecutive days followed by 3 days of no-stimulations to accommodate the treatment cycles of TMZ (see below). Sham controls underwent the same handling procedure as the kindled groups except no current was delivered.

The behavioral convulsion induced after each stimulation was recorded using a modified Racine scale (Racine, 1972b): Class 0 – freezing; Class 1 – orofacial automatisms; Class 2 – orofacial automatisms with head nodding; Class 3 –unilateral forelimb clonus; Class 4 – rearing with bilateral forelimb clonus; Class 5 – rearing with bilateral forelimb clonus followed by falling; Class 6 – multiple Class 5′s. Rats were returned to their home cage once all convulsive behaviors subsided. Rats were considered to be “kindled” after achieving three consecutive Class 5 or higher convulsions.

Temozolomide (TMZ) has been widely used as a pharmacological approach to investigate the effects of reduced adult neurogenesis on behavior in mice and rats. TMZ (MedChem Express, CAS: 825622-93) is a DNA alkylating prodrug with high blood-brain barrier permeability that is commonly used as an oral treatment for high-grade glioblastoma multiforme. TMZ (2.5 mg/ml) was freshly prepared each day by dissolving in ice-cold sterile 0.9% (w/v) saline (pH 5.5–6.0). The TMZ solution was then vortexed (~30 s) and briefly sonicated (50 W, 5 s) in an ice bath. The TMZ solution was sterile filtered before use to remove remaining insoluble residues and remained on ice during injections.

In this study, rats were administered twice-daily injections of TMZ (25 mg/kg, i.p.) or saline, with injections spaced at least 4 h apart, for three consecutive days with a 4-day kindling period immediately following the last TMZ treatment. The first cycle of TMZ treatment for the kindled groups began after the 30th or 45th electrical stimulation to target the period of enhanced neurogenesis during kindling. Subsequent treatment cycles continued until the animal received a total of either 75 or 105 stimulations (Figures 1A, B). This ensured that both TMZ and saline-treated kindled groups achieved a similar number of elicited motor convulsions during the kindling procedure. During the treatment free period, kindled and sham stimulations (3 stimulations per day) resumed over the next 5 days until the following cycle of TMZ or saline began. Kindled rats received 4 or 5 treatment cycles of TMZ or saline during kindling period. To minimize potential confounding issues related to TMZ treatment and/or indirect (postictal) effects from kindling on behavior, a 1-week recovery or rest period was given. Using this protocol, TMZ has been demonstrated to reliably inhibit adult neurogenesis without causing general impairments in the health of the animal (Garthe et al., 2009; Nokia et al., 2012; Bambico et al., 2015; Perez-Domper et al., 2017; Nickell et al., 2020).

To assess for changes in levels of adult hippocampal neurogenesis after kindling, we examined the immature neuronal marker doublecortin (DCX). DCX is a microtubule-associated protein associated with neuronal differentiation and neurite elongation, which is exclusively expressed in immature dentate granule cells from 1 day to about 4 weeks post-mitosis (Brown et al., 2003; Plumpe et al., 2006). As expected, DCX-immunolabeled cells were found predominately clustered along the border of the dentate granule cell layer and subgranular zone (Figure 2A). Using unbiased stereological cell counting procedures, we found a significant effect of group on the total number of DCX+ cells in the dentate gyrus [ANOVA, F(2, 57) = 23.98, P < 0.001, η² = 0.46]. The results presented in Figure 2 show that kindled rats had an almost 1.7-fold increase in the number of DCX+ compared to sham controls (Sidak, P < 0.001). As shown in Figure 2B, multicycles of TMZ treatment markedly suppressed the effect of kindling on hippocampal neurogenesis as the number of DCX+ cells in this group was significantly lower than rats that received kindling only (Sidak, P < 0.002) and approached the levels of sham controls (Sidak, P = 0.788).

Given that the dorsal (septal) and ventral (temporal) hippocampal regions are anatomically and functionally distinct, we also examined if kindling might differentially affect neurogenesis across the septotemporal axis of the dentate gyrus. Compared to sham controls, the mean area fraction of DCX immunoreactivity was increased by 3.3-fold in the dorsal dentate gyrus (Figure 2C) and ~2-fold in the ventral dentate gyrus (Figure 2C) after kindling [Dorsal: two-way ANOVA, F(2, 57) = 28.3, P < 0.001, η² = 0.49, ω² = 0.47; Ventral: two-way ANOVA, F(2, 57) = 14.45, P < 0.0001, η² = 0.33, ω² = 0.31]. However, the kindling-induced increase in DCX immunoreactivity was significantly attenuated by TMZ treatment. Although TMZ reduced overall decrease in DCX levels across the dorsal and ventral dentate gyrus after kindling, this reduction was more pronounced in the ventral dentate gyrus (kindled/TMZ: dorsal vs. ventral, paired t-tests, P = 0.009), where the mean area fraction of DCX labeling was no longer significantly different from that of sham controls. In contrast, DCX immunoreactivity in the dorsal dentate gyrus remained slightly elevated for the TMZ group compared to sham controls, though this difference only approached statistical significance [P = 0.059]. Taken together, our findings suggest that cyclic treatment of TMZ (beginning after the 30th or 45th stimulation) can effectively suppress kindling-induced hippocampal neurogenesis.

Seizures not only enhance the generation of new neurons but can also promote aberrant morphological development and migration of these cells. DCX+ cells in the non-kindled controls had typical granular morphology and generally displayed only a single apical dendritic process that reached the molecular layer before ramifying into additional smaller branches. However, the morphology of DCX+ cells from kindled rats displayed more irregular-shapes, characterized by flattened and elliptical-like cell bodies, and showed more expansive dendritic growth and branching within molecular layer. Remarkably, many of these morphological changes were reduced in kindled rats that received TMZ. While there was also some DCX+ cells that ectopically migrated into the hilus after kindling [ANOVA, F(2, 57) = 7.13, P < 002, η² = 0.20, ω² = 0.17, Figure 2D], this was almost completely suppressed after TMZ treatment. Together these results suggest that kindling promotes an increase in aberrant neurogenesis that can be prevented by multicycle treatment with TMZ.

To investigate if blocking the increase in aberrant hippocampal neurogenesis that occurred with kindling could prevent learning deficits, animals underwent training in a contextual pattern separation task (see Figure 1C for task procedure). This task required the animal to discriminate between two highly similar conditioning contexts: a shock-paired context (A) and a slightly modified context that was never paired with a foot shock (A'). We performed a three-way mixed design ANOVA with Day (1 through 6) and Context (A vs. A') as the within subject factors and Group (sham vs. kindled vs. kindled/TMZ) as the between subject factor to examine for changes in freezing across the conditioning sessions. The results revealed a significant Group by Context interaction [RM-ANOVA, F(2, 50) = 8.02, P < 0.001, partial η² = 0.24] and a marginally significant three-way interaction involving Group, Day, and Context [RM-ANOVA, F(10, 250) = 1.89, P = 0.077, partial η² = 0.07]. Significant main effects for Context [F(1, 57) = 9.49, P < 0.003, partial η² = 0.16], Day [F(5, 286) = 37.74, P < 0.001, partial η² = 0.43], and Group [F(2, 50) = 7.67, P < 0.001, partial η² = 0.23] were also found.

For context A (the shock only context), all groups showed equivocal levels of freezing to this context over the 6 days of conditioning (RM-ANOVA, All Fs < 1.88, All Ps > 0.162). In contrast, analysis of context A' (the no shock context) found group differences in freezing across the conditioning sessions [RM-ANOVA, Group by Day, F(10, 285) = 2.87, P < 0.002, partial η² = 0.10, Figure 3A]. As shown in Figure 3B, kindled animals irrespective of treatment displayed initially higher freezing scores to context A' than the sham controls on the first day of conditioning (Sidak, P < 0.015). However, by the following day, all groups showed similar freezing to context A'. Over the remaining conditioning days, kindled rats continued to show higher levels of freezing to context A' (the no shock) compared to sham controls, however, this difference was only significant on Days 5 and 6 of training when contrasted with kindled rats that received TMZ (Sidak All Ps < .026). Importantly, there was no difference in freezing to the no shock context (A') for sham controls and kindled rats that received TMZ over the remainder of these conditioning sessions suggesting that both groups demonstrated similar levels of freezing to this context.

To further examine if kindled rats had difficulty distinguishing between contexts A and A' throughout all conditioning sessions, we conducted a series of paired t-tests to determine when significant differences in freezing to the two contexts emerged. Sham controls showed a significant decrease in freezing to context A' compared to context A on day 3. This difference persisted and remained statistically significant until day 6 (paired t-tests, All Ps < 0.019, All Cohen's ds = 0.497 to 1.17, Day 6 - Figure 3C). In contrast, kindled rats that received saline displayed high levels of freezing to both context A and A' across all training session indicting an overall difficulty in discriminating between the two contexts. Interestingly, treatment with TMZ appeared to mitigate the effect of kindling. Similar to sham controls, kindled rats that received TMZ showed reduced freezing to the no-shock context A' when contrasted with the shock-paired context A with this difference beginning one day after sham controls (Day 4, paired t-test, P < .038, Cohen's d = 0.51) and remaining until the end of conditioning (paired t-tests, All Ps < 0.013, Cohen's ds = 0.55 to 0.74, Day 6 – Figure 3C). These finding suggest that kindled rats may have difficulties discriminating between the two contexts leading to increased fear generalization. However, chemical ablation of neurogenesis during kindling appears to alleviate this deficit resulting in improved contextual discrimination learning.

To examine if blocking seizure-induced neurogenesis might impact performance on another behavioral task, a subset of animals underwent testing for novel object recognition (sham n = 11; Kindled n = 10; Kindled/TMZ n = 10). Following a 30-minute retention interval, a one-sample t-test was used to examine whether the discrimination index significantly differed from 50% (chance level). As shown in Figure 4, both sham controls and the kindled rats received TMZ appeared to successfully discriminate the novel object from the familiar object during the recall test [one-sample t-tests, sham, t(8) = 3.77, P < 0.005, Cohen's d = 1.26; kindled/TMZ, t(9) = 2.73, P < 0.023, Cohen's d = 0.86]. However, the discrimination index for kindled group that received saline did not differ significantly from chance [one-sample t-test, t(9) = 1.51, P = 0.155, Cohen's d = 0.42].

We confirm past work (Parent et al., 1998; Scott et al., 1998; Fournier et al., 2010, 2013; Chen et al., 2023) that electrical kindling of the amygdala dramatically increases the number of immature neurons (DCX+ cells) generated in the dentate gyrus. Although birth-dating experiments with the thymidine analog BrdU was not performed, we have previously shown that the increased proliferation or birth of new neurons tends to be highest during the early stages of electrical kindling—beginning round 30th stimulation and peaking at the 45th stimulation—when rats begin to regularly display overt motor convulsions in response to stimulation and meet the criteria for being classified as “kindled” (Fournier et al., 2010, 2013). During this early stage of kindling, heightened network stimulation appears to drive neural stem cell activity leading to a rapid expansion of proliferating pools of neuroprogenitor cells. However, as kindling stimulations extend beyond this point, repeated evocation of seizures lead to changes in the morphology and connectivity of adult-born neurons, along with a decline in proliferation and neurogenesis toward baseline values. For example, in both this study and our previous work (Fournier et al., 2010, 2013) we found that several DCX+ cells exhibit migration deficits with displacement into the hilus. In addition, several DCX+ cells located in in the granule cell layer and subgranular zone displayed aberrant structural alterations, including formation of hilar basal dendrites and increased dendritic branching. These findings are reminiscent of what has been commonly reported in multiple rodent chemoconvulsant models of status epilepticus, such as pilocarpine and kainic acid (Scharfman et al., 2000; Hattiangady et al., 2004; Jessberger et al., 2007; Walter et al., 2007; Pierce et al., 2011; Hester and Danzer, 2013; Jain et al., 2019) suggesting that seizure stimulation—despite being evoked through different experimental methods—appear to produce similar disruptions in proliferation, maturation, and connectivity of newborn neurons.

Several lines of evidence show that adult-born neurons (between 2 to 6 weeks of age postmitosis) exhibit greater intrinsic excitability and plasticity than their mature counterparts (Ash et al., 2023). As a result, immature neurons at this stage of development may be especially vulnerable to the effects of pathological stimulation, such as seizure activity. In support of this, Lybrand et al. (2021) and others (Singh et al., 2015; Hosford et al., 2016; Althaus et al., 2019; Zhou et al., 2019; Chen et al., 2023) found that when seizure activity occurred throughout this critical window it drove ectopic migration, aberrant morphological development and accelerated integration of new neurons. This window overlaps with the period of extended kindling in our study and aligns with our finding that repeated seizures evoked during this time can disrupt the normal development of new neurons.

We hypothesized that TMZ treatment during this critical period might be an effective strategy to reduce aberrant neurogenesis and improve hippocampal function. Our findings support this. First, we found that TMZ significantly blunted kindling-evoked neurogenesis, with the number of DCX+ cells being comparable to sham controls at the end of kindling. Second, TMZ also prevented the generation of aberrant newly generated neurons as there was an almost complete suppression of ectopically displaced DCX+ cells into the hilus along with a reduction in DCX+ cells that exhibited abnormal morphology. These findings suggest that most aberrant new neurons born during kindling appear to be generated during this early wave of enhanced neural stem cell/progenitor activity.

Although TMZ was effective in arresting kindling-induced neurogenesis across the septotemporal axis of the hippocampus, this effect appeared greater in the ventral hippocampus compared to the dorsal hippocampus. The exact reason for this difference is unclear. However, Egeland et al. (2017) reported in mice that 6-week of TMZ treatment caused a greater reduction in neurogenesis in the ventral hippocampus compared to the dorsal hippocampus suggesting that the ventral hippocampus may be more vulnerable to the effects of antimitogenic agents such as TMZ. Since the ventral hippocampus displays overall lower rates of neurogenesis and proliferative activity than the dorsal hippocampus (Wu et al., 2015), it is possible that these differences could enable the impact of TMZ on neurogenesis to be more apparent. In fact, we found that while TMZ reduced overall levels of neurogenesis in the dorsal and ventral hippocampus after kindling, DCX+ immunolabelling still trended toward being significantly elevated within the dorsal hippocampus compared to sham controls. Regardless, our findings confirm the anti-mitogenic and suppressive actions of TMZ on hippocampal neurogenesis as previously reported in multiple rodent studies further highlighting its usefulness as a tool to study neurogenesis in the adult brain (Garthe et al., 2009; Egeland et al., 2017; Che et al., 2021).

The accurate migration, development, and integration of new neurons is crucial for hippocampal function. Thus, a central question of this study was to investigate whether the ongoing production of abnormally integrated new neurons during kindling could explain, at least in part, the development of cognitive and behavioral deficits that are seen in chronic epilepsy. To address this, we used a contextual fear discrimination task which requires the animal to distinguish between a context associated with a mild footshock and a similar context that was never paired with the shock. Contextual discrimination has been shown to depend on pattern separation and correlates with levels of intact adult hippocampal neurogenesis (Sahay et al., 2011; Niibori et al., 2012; Tronel et al., 2012; Winocur et al., 2012; Denny et al., 2014; Miller and Sahay, 2019).

Our findings clearly show that repeated seizures induced by electrical kindling caused pronounced deficits in contextual discrimination learning. For example, while non-kindled controls could effectively discriminate between the shock and non-shock contexts by Day 3 of training, kindled rats were unable to do so and displayed high levels of freezing to both contexts across the training sessions. Enhanced fear generalization has been linked to behavioral pattern separation deficits and suggests repeated kindled seizures may sensitize neural circuits involved in threat detection and defensive responses. This parallels clinical observation of elevated interictal anxiety and fear among patients with mesial temporal lobe epilepsy (Hermann et al., 1982; Heilman, 2021; Rauh et al., 2022). In fact, we have previously shown that long-term amygdala kindling can also produce exaggerated anxiety and enhanced fear generalization in rodents (Kalynchuk and Fournier, 2009; Fournier et al., 2020).

Most importantly, we found that impairments in context discrimination were ameliorated by TMZ treatment. Kindled rats that received TMZ showed discriminative freezing to both contexts by Day 4 of training, indicating that inhibition of seizure-induced neurogenesis appears to mitigate discriminative fear deficits produced by kindling. In this context, our finding that TMZ appeared to produce greater suppression of kindling-evoked neurogenesis within the ventral hippocampus might be particularly relevant. The ventral hippocampus is known to be highly implicated in emotional regulation and in the disambiguation between threatening and safe contexts (Fournier and Duman, 2013; Kheirbek et al., 2013; Besnard et al., 2020). It is possible that aberrant neurogenesis within the ventral hippocampus could disrupt the balance between excitatory and inhibitory circuits which could lead to overgeneralized fear responses (Li et al., 2024). While speculative, our findings suggest that by reducing seizure-induced neurogenesis, particularly within the ventral hippocampus, TMZ may help restore circuit stability which could improve pattern separation and reduce inappropriate fear generalization across contexts (Besnard and Sahay, 2016). In addition to this, we also observed a protective effect of TMZ on another behavioral task, the novel object recognition test. We found that kindled rats treated with TMZ demonstrated comparable levels of recall as sham controls during testing, whereas non-treated kindled rats showed impairments in recognition memory. Object recognition is typically associated with extra-hippocampal areas, such as the perirhinal cortex. This suggests that TMZ preservation of object recognition memory may extend beyond its effect on suppressing seizure-induced neurogenesis. However, object-based and novelty-related information is still relayed to the hippocampus to support overall object processing (Kinnavane et al., 2016). Therefore, by preventing seizure-induced neurogenesis within the hippocampus, TMZ may help offset the formation of aberrant network changes that disrupt interactions between the hippocampus and other brain regions, such as the perirhinal cortex, during object learning. Taken together, our findings highlight an important role of seizure-induced neurogenesis as a potential mechanism underlying the development of epilepsy-associated cognitive impairments, and that targeting this process through chemical ablation with TMZ can prevent cognitive decline associated with chronic seizures.