Interictal spike frequency varies with ovarian cycle stage in a rat model of epilepsy

Abstract

In catamenial epilepsy, seizures exhibit a cyclic pattern that parallels the menstrual cycle. Many studies suggest that catamenial seizures are caused by fluctuations in gonadal hormones during the menstrual cycle, but this has been difficult to study in rodent models of epilepsy because the ovarian cycle in rodents, called the estrous cycle, is disrupted by severe seizures. Thus, when epilepsy is severe, estrous cycles become irregular or stop. Therefore, we modified kainic acid (KA)- and pilocarpine-induced status epilepticus (SE) models of epilepsy so that seizures were rare for the first months after SE, and conducted video-EEG during this time. The results showed that interictal spikes (IIS) occurred intermittently. All rats with regular 4-day estrous cycles had IIS that waxed and waned with the estrous cycle. The association between the estrous cycle and IIS was strong: if the estrous cycles became irregular transiently, IIS frequency also became irregular, and when the estrous cycle resumed its 4-day pattern, IIS frequency did also. Furthermore, when rats were ovariectomized, or males were recorded, IIS frequency did not show a 4-day pattern. Systemic administration of an estrogen receptor antagonist stopped the estrous cycle transiently, accompanied by transient irregularity of the IIS pattern. Eventually all animals developed severe, frequent seizures and at that time both the estrous cycle and the IIS became irregular. We conclude that the estrous cycle entrains IIS in the modified KA and pilocarpine SE models of epilepsy. The data suggest that the ovarian cycle influences more aspects of epilepsy than seizure susceptibility.

Keywords: Animal model; Epilepsy; Hormone; Kainic acid; Neuropathology; Seizure; Women.

Figure 1. Kainic acid (KA)- and pilocarpine-induced SE in female rats and its use to study the relationship between the estrous cycle and the EEG

A. 1–2. KA injection led to a high incidence of SE. Black: animals with SE; grey: animals that had few or no seizures (i.e., they did not have SE); white: animals that died. 3. Incidence of SE at different estrous cycle stages. Numbers above the bars are (numbers of animals with SE)/(total number of animals tested). There were no significant differences (Chi-squared test, p>0.05). B. Experimental procedures. 1. Females with regular estrous cycles were injected with pilocarpine or KA and electrodes were implanted approximately 1 month later. After 1 week for recovery, animals were recorded daily in the morning and vaginal cytology was used to confirm that estrous cycles were maintained. Mon = month. 2. Ovx females had approximately 1 week for recovery and then were recorded daily. 3. Males were treated like intact females except there was no cytologic examination. 4. Pharmacology followed the same procedures as in D1, except for drug injections at the arrows. The drug injections were made prior to the recording session of that day. 5. A cohort of animals was treated the same as for #1 but were perfusion-fixed 3 days after KA injection.

Figure 2. Example of a female rat after KA-induced SE with an increase in IIS frequency on metestrous and diestrous 2 mornings

A. 1. A sample of the EEG during exploration on proestrous morning shows no IIS. LOC, left occipital cortex; LFC, left frontal cortex; LHC, left hippocampus; RHC, right hippocampus; RFC, right frontal cortex, ROC, right occipital cortex. 2–3. The EEG recordings marked by the bars are expanded as noted by the arrows. B. The same animal, recorded one day later (on estrous morning). The expanded EEG shows a single, very small spike that was recorded primarily in hippocampal channels and is therefore likely to be a sharp wave. C. One day later, metestrous morning, numerous spikes are present in the EEG. Expansion of the record in 2–3 shows the large size and duration of IIS that distinguished it from the spike in B3. D. One day later, diestrous 2 morning, many IIS are also present. The expanded records in D2-3 show that the IIS are similar although smaller and less complex than the previous day, suggesting that IIS frequency as well as amplitude and complexity changed across the estrous cycle. In E, frequency is quantified rather than amplitude and complexity because the latter varies from channel to channel and animal to animal, with a large component of the variance due to electrode location. E. Mean IIS frequency is plotted for 4 consecutive estrous cycles for exploration (blue bars) and behavioral arrest (green bars). A two-way ANOVA with cycle stage and behavioral state as factors showed a significant effect of cycle stage [F(3,32)4.953, p=0.0062] with no interactions of factors [F(3,32)0.384, p=0.7650]. Cycle stages were significantly different for exploration [one-way ANOVA, F(3,16)3.920, p=0.0284] but not behavioral arrest [F(3,16)1.525, p=0.2463]. For exploration, all pairwise comparisons were significant (Neuman-Keuls multiple range test, asterisks, p<0.05) except metestrus vs. diestrus 3, and proestrus vs. estrus.

Figure 3. Example of a female rat after KA-induced SE with an increase in IIS frequency on estrous and metestrous mornings

A. 1. A sample of the EEG during exploration shows no IIS on proestrous morning. 2–3. The parts of the EEG records in A1 that are marked by the bars are expanded. B. The same animal, recorded 24 hrs later (estrous morning) shows numerous IIS. C–D. During metestrous and diestrous 2 mornings, IIS were not evident. Note that IIS sometimes occurred on metestrous morning in this animal, but did not occur during this example. The variance is evident in the relatively large standard error bar for metestrous morning (M) in E for exploration. E. Mean IIS frequency is plotted for 4 consecutive estrous cycles. IIS recorded during exploration (blue bars), behavioral arrest (green), and sleep (gray) show a similar predisposition for estrous and metestrous mornings. A two-way ANOVA with cycle stage and behavioral state as factors showed a significant effect of cycle stage [F(3,36)16.669, p<0.001] with no interactions of factors [F(6,36)1.365, p=0.2550]. For exploration, there was a significant effect of cycle stage by one-way ANOVA [(F(3,12)20.673; p<0.0001)]. Cycle stage was also a significant factor for behavioral arrest [one-way ANOVA: (F(3,12)3.534, p=0.0484)] and sleep [one-way ANOVA: F(3,12)3.532, p=0.0485]. For all behavioral states, the pairwise comparison between diestrus 2 and estrus was significant (Neuman-Keuls multiple range test, p<0.05). For exploration, additional pairwise comparisons were also significant as indicated by the brackets and asterisks (all p<0.05). M= metestrous morning; D= diestrous 2 morning; P= proestrous morning; E= estrous morning.

Figure 4. In KA-treated female rats with regular estrous cycles fluctuations in IIS frequency are cyclic and similar in periodicity to the estrous cycle

A. IIS frequency is plotted for consecutive estrous cycles with recordings during exploration (1) or behavioral arrest (2). These data were obtained from a single rat and are shown to exemplify variation in IIS frequency in relation to the estrous cycle. M= metestrous morning; D= diestrous 2 morning; P= proestrous morning; E= estrous morning. B. Sine wave regression analyses for two different animals (one in B1, the other in B2) are shown. In each case, the raw data (IIS/min) for exploration (blue lines) and behavioral arrest (green lines) fit a sine wave pattern (red lines). The r and p values are listed in the upper left corners of each graph.

Figure 5. Acyclicity reduces the cyclic pattern of IIS frequency after KA-induced SE

A. IIS frequency is plotted for an animal that had a spontaneous period where there was acyclicity according to the vaginal cytologic data (dotted line). Prior to the acyclicity, IIS frequency increased on proestrous (P) and estrous (E) mornings. During the time when vaginal cycles were acyclic, IIS became acyclic. An arrow points to the first day when IIS frequency did not follow the pattern of increasing in frequency on proestrous or estrous morning. This day occurred prior to the onset of vaginal acyclicity (i.e., defined by vaginal cytology), presumably because the turnover of cells in the vaginal epithelium is 24–36 hrs slower than the changes in the brain. Metestrus, M; diestrus 2, D. B. Data from the same animal during behavioral arrest. During the period of vaginal acyclicity, IIS frequency decreased (the period of acyclicity corresponds to the dotted line on the X axis). Thus, acyclicity disrupted the cyclic IIS fluctuations and led to a dissociation between IIS in exploration (increasing during acyclicity) and behavioral arrest (decreasing). C.1. The data for behavioral arrest in B are plotted after sine wave regression analysis to show that the sinusoidal pattern that occurred when the animal had regular vaginal cycles was decreased during acyclicity. In C1 and C2, the acyclic period is indicated by a dotted line. 2. Data from another rat showing a dramatic reduction in IIS when there was acyclicity. The recordings were made during behavioral arrest.

Figure 6. Ovx reduces the cyclic pattern of IIS frequency after KA-induced SE

A–B. 1. IIS are shown for a day when IIS were robust before (A, Pre-Ovx) and 7 days after (B, Post) Ovx. 2. Parts of the EEG recordings in A1 that are marked by a bar are expanded. 3. Parts of the EEG recordings in A2 that are marked by a bar are expanded. C. Mean IIS frequency is plotted for 12 consecutive days. The first 4 days were numbered 1–4, the next 4 days as 1–4, and the last 4 days as 1–4. Days were assigned a number to test the hypothesis that a 4-day sequence was present. All days assigned 1 were then averaged, those days numbered 2 were averaged, and means for days 3 and 4 were also calculated. Two-way ANOVA, with the day (i.e., 1,2,3, or 4) or behavioral state as factors showed no influence of day [F(3,56)1.396, p=0.2535] but there was a difference in behavioral state [F(2,56) 5.076, p=0.0094], consistent with a dissociation of IIS frequency for exploration and behavioral arrest after Ovx. Thus, there were relatively high IIS frequencies during exploration in Ovx rats but relatively low frequencies during behavioral arrest. IIS were particularly infrequent during sleep. There was no interaction of factors [F(6, 56) 0.699, p=0.6511].

Figure 7. Pharmacological cessation of the estrous cycle reduces cyclic IIS

IIS frequency during exploration is shown for consecutive days of a rat that had KA-induced SE approximately 3 months earlier. IIS frequency increased on estrous and metestrous mornings in this animal, highlighted in yellow. Two vehicle (black arrows) or two ICI 182,780 (red arrows) injections were made in the early morning of diestrous 2 and proestrous. The injections of ICI were intended to block nuclear estrogen receptors and therefore the actions of estrogen that are critical to ovulation late on proestrous day. Vaginal cytology showed that the estrous cycle had an abnormal pattern on the days after the injection of ICI 182,780 (indicated by a dotted line) but the effect was not detected after vehicle treatment. IIS frequency was low during the period of vaginal acyclicity. Afterwards the estrous cycle resumed and there was a return of the pattern in IIS frequency to the one that was observed before injection of ICI, an increase in IIS frequency during estrous and metestrous mornings.

Figure 8. Males lack a 4-day IIS frequency pattern after KA-induced SE

A–D. 1. EEG of a male, 8 weeks after KA-induced SE. Data from 4 consecutive mornings are shown. During these recordings, animals were exploring. IIS occurred each day. 2. Areas of the EEG recordings that are marked by the bars in A–D are expanded. E. 1–2. Quantification of IIS frequency as in Figure 6, i.e., the first 4 days assigned the numbers 1–4, the next 4 days also assigned numbers 1–4, etc. The means for all of the days assigned 1, 2, 3, or 4 were then compared by two-way ANOVA with day and behavioral state as factors. There was no influence of day [F(3,26)0.794, p=0.5081] or behavioral state [F(1,26)0.063, p=0.8038].

Figure 9. Cyclic IIS occur in female rats after pilocarpine-induced SE, like KA-induced SE

A–D. 1. An example of a pilocarpine-treated rat that had SE, regular estrous cycles, and cyclic IIS. 2. The parts of the EEG recordings in A–D that are marked by bars are expanded. Note that there are sharp waves in hippocampus but these are not accompanied by spikes in frontal cortex or occipital cortex, which distinguishes sharp waves from IIS. E. Mean IIS frequencies for 4 consecutive estrous cycles are plotted for exploration, behavioral arrest and sleep. Two-way ANOVA showed that there was an effect of cycle stage [F(3,36)7.770, p=0.004] and behavioral state [F(2,36)3.264, p=0.0498) and no interaction of factors [F(6,36) 0.451, p=0.8392). For exploration, one-way ANOVA followed by pairwise comparisons showed significant differences between all cycle stages except metestrus and diestrus, as indicated by the asterisks (Neuman-Keuls multiple range tests, all p<0.05). There were no significant differences between cycle stages for behavioral arrest and sleep.

Figure 10. KA-induced SE on diestrus led to the least neuronal damage in female rats

A. Sections from 3 animals that were perfused 3 days after KA-induced SE were stained using an antibody to NeuN. Extensive NeuN loss was evident in the male (top, arrows), female that had SE on proestrous morning (center), but not the female that had SE on metestrous morning (bottom). Calibrations for A and B are the same. 1. Hippocampal subfields showing a loss of NeuN are marked by arrows and include the hilus (HIL), CA3, CA1, and the subiculum (Sub). 2. Areas showing loss of NeuN in the entorhinal cortex (EC) were located in the medial and lateral divisions (MEC, LEC). Pre, presubiculum; para, parasubiculum. 3. Quantification of NeuN loss. Each circle represents data from a different animal, which were injected with KA at different cycle stages or were males. Left: The area fraction of NeuN loss showed sparing of the granule cell layer (GCL, close to 0 %). There was NeuN loss in the CA1–CA3 cell layer (CA1–CA3), hilus (HIL) and subiculum (SUB) for all animals. Yellow, SE on metestrus or diestrus 2 morning (n=4); Red, SE on proestrous or estrous morning (n=3); Black, males (n=7). Right: The area fraction (%) is plotted for the pre and parasubiculum (Pre + Para), all layers of the MEC, and just layer III of the MEC. Note layer III of the MEC exhibited the most NeuN loss of all animals, which has been shown for male rats that have SE [50] B. Adjacent sections to those in A were stained for fluorojade B. The areas corresponding to NeuN loss in A showed positive staining with fluorojade B in most cases, but there were exceptions. For the hilus and entorhinal cortex, females that had SE during diestrus showed weak fluorojade B staining. In these cases, there was damage to neurons (pkynosis using cresyl violet staining) but not death (loss of cells by cresyl violet staining), consistent with a loss of NeuN in response to injury (36). B. Fluorojade staining in the same animals as A1-2. 1. Hippocampus. 2. Entorhinal cortex. 3. The area surrounded by the box in (2) is expanded to show the fluorojade B-positive cells in the deep and superficial layers of the EC. LD, lamina dissecans. Note that there is extensive fluorojade B staining except for the female that had SE on metestrous morning where fluorojade B cells were evident but only in the most medial part of the EC (arrows). 4. Quantification of fluorojade B-stained cells. Females had similar staining as males except for the females that had SE during diestrus. Because females that had SE during diestrus were the animals which maintained estrous cycles and exhibited cyclic IIS, the data suggest that only those animals without extensive neuronal loss had preservation of their estrous cycles and developed cyclic IIS.