Seizure-induced formation of isofurans: novel products of lipid peroxidation whose formation is positively modulated by oxygen tension


Address correspondence and reprint requests to Manisha Patel, Department of Pharmaceutical Sciences, University of Colorado Health Sciences Center, Box C238, Denver, CO 80262, USA.


We have previously shown that seizures induce the formation of F2-isoprostanes (F2-IsoPs), one of the most reliable indices of oxidative stress in vivo. Isofurans (IsoFs) are novel products of lipid peroxidation whose formation is favored by high oxygen tensions. In contrast, high oxygen tensions suppress the formation of F2-IsoPs. The present study determined seizure-induced formation of IsoFs and its relationship with cellular oxygen levels (pO2). Status epilepticus (SE) resulted in F2-IsoP and IsoF formation, with overlapping but distinct time courses in hippocampal subregions. IsoF, but not F2-IsoP formation coincided with mitochondrial oxidative stress. SE resulted in a transient decrease in hippocampal pO2 measured by in vivo electron paramagnetic resonance oximetry suggesting an early phase of seizure-induced hypoxia. Seizure-induced F2-IsoP formation coincided with the peak hypoxia phase, whereas IsoF formation coincided with the ‘reoxygenation’ phase. These results demonstrate seizure-induced increase in IsoF formation and its correlation with changes in hippocampal pO2 and mitochondrial dysfunction.

Abbreviations used

dentate gyrus


electron paramagnetic resonance






lithium phthalocyanine




reactive oxygen species


status epilepticus

Free radicals have been implicated in the pathogenesis of many neuronal diseases; however, their role in epilepsies is only recently beginning to be recognized (Kovacs et al. 2002; Kunz 2002; Patel 2004). The selective oxidation of susceptible cellular macromolecules to attack renders them suitable as surrogate markers of oxidative stress in vivo. Assessment of seizure-induced oxidative damage to susceptible subcellular targets of oxidative damage (protein, lipids, and DNA) and redox status, suggests that status epilepticus (SE) produces profound mitochondrial oxidative stress (Patel 2004). Additionally, dramatic metabolic and bioenergetic changes occur as a consequence of both acute seizure episodes and chronic epilepsy (Fernandes et al. 1999; Hetherington et al. 2002; Meldrum 2002). The acute consequence of SE is an increase in cellular glucose uptake and metabolism that is unparalleled by most conditions (Meldrum 1983; Theodore 1999; Cornford et al. 2002). Despite an increase in cerebral blood flow to match this hypermetabolism (Meldrum 1983; Franck et al. 1986; Pereira de Vasconcelos et al. 2002), a disparity between high energy demand and brain oxygenation occurs (Kreisman et al. 1984).

There is convincing evidence that seizure activity increases the levels of free fatty acids and diacylglycerol and the formation of arachidonic acid (Bazan et al. 1986). Prolonged seizures produce a large increase in prostaglandins derivatives including prostaglandin-F2α (Bazan et al. 1986). Seizure-induced lipid peroxidation has been observed by measuring products such as thiobarbituric acid reactive substances (Bruce and Baudry 1995) and F2-isoprostanes (F2-IsoPs) (Roberts and Morrow 2000; Patel et al. 2001). F2-IsoPs are a novel class of prostaglandin F2-like compounds, produced in vivo by a non-cyclooxygenase and free radical-catalyzed mechanism involving the peroxidation of arachidonic acid (Morrow et al. 1990). F2-IsoPs serve a sensitive, stable, and reliable marker of free radical-induced lipid peroxidation in vivo. We have previously demonstrated seizure-induced formation of F2-IsoPs in hippocampal subregions (Patel et al. 2001). Recently, novel products of free radical-mediated peroxidation of arachidonic acid were discovered that have a substituted tetrahydrofuran ring, which have been termed isofurans (IsoFs). Interestingly, it was shown that the formation of IsoFs and F2-IsoPs are differentially modulated by oxygen tension; high oxygen tension favors the formation of IsoFs whereas the formation of F2-IsoPs is favored at low oxygen tension (Fessel et al. 2002). As SE may produce local changes in tissue oxygen tension, we examined whether prolonged seizures increase IsoFs and if so how their formation correlates with mitochondrial oxidative stress and changes in tissue oxygen tension measured by in vivo electron paramagnetic resonance (EPR) oximetry.

Materials and methods

Animal studies with the exception of tissue pO2 analysis by EPR were conducted at the University of Colorado Health Sciences Center in accordance with institutional regulations. The animal study involving pO2 analysis by EPR was conducted at Dartmouth College in accordance with their institutional policies. Efforts were made to minimize animal suffering and to reduce the number of animals used. Tissue analysis of F2-IsoP and IsoF was conducted at Vanderbilt University.

Kainate administration

Adult male Sprague–Dawley rats (200–250 g) were injected with saline or kainic acid (Bio Vectra, Charlottetown, PE, Canada) dissolved in saline (12 mg/kg, s.c.). The severity of behavioral seizures was evaluated continuously (every 30 min during the first 4 h and hourly thereafter) during a 8-h period following kainate treatment according to the following rating scale adapted from (Ben-Ari 1985; Baran et al. 1987) and previously described (Liang et al. 2000).

Microdissection of hippocampal subregions

Transverse hippocampal slices (600-μm thick) were prepared using a McIlwain tissue chopper (Mickle Laboratory Engineering Co., Guilford, Surrey, UK). Slices were immersed in phosphate-buffered saline containing 0.001% butylated hydroxytoluene to prevent further oxidation and in situ formation of F2-Isops or IsoF. Using an illuminated dissected microscope, the hippocampal formation was dissected into three parts, the CA3-CA2 area (designated as CA3), CA1 region, and dentate gyrus including part of the CA4 region (designated as DG). The tissue was collected in chilled pre-weighed tubes placed on dry ice until analysis as described below.

F2-Isop and IsoF measurement

The measurement of F2-IsoPs and IsoF in the frozen samples was conducted at Vanderbilt University such that identity of the treatment groups was disclosed following the analysis. Both F2-IsoPs and IsoFs esterified in tissue lipids were quantified by gas chromatography negative ion chemical ionization mass spectrometry in a single assay as previously described (Morrow and Roberts 1999; Fessel et al. 2002). Briefly, lipids were extracted from tissue by the Folch method (CHCl3/methanol 2 : 1, v/v), containing 0.005% butylated hydroxytoluene. The lipids were evaporated to dryness and hydrolyzed with KOH (15%) to release free F2-IsoPs and IsoFs. Free F2-IsoPs and IsoFs were then extracted using a C18 Sep-Pak column, converted to pentafluorobenzyl esters, purified by thin layer chromatography, derivatized to trimethylsislyl ether derivatives, and quantified by gas chromatography negative ion chemical ionization mass spectrometry using (2H4) 15-F2t-IsoP as an internal standard. Values are expressed as nanograms of F2-IsoPs or IsoFs per gram tissue.

Mitochondrial isolation and enzyme activities

Preparation of mitochondrial fractions from microdissected hippocampal regions was conducted as described previously (Liang et al. 2000). Purity of the mitochondrial and cytosolic fractions was determined by measuring cytochrome oxidase and lactate dehydrogenase activities, respectively. Lactate dehydrogenase activity was on average 5-fold higher in cytosolic fractions compared with mitochondrial fractions. Cytochrome oxidase, assessed by enzymatic and immunoblot analysis was virtually undetectable in cytosolic fractions and robustly expressed in mitochondrial fraction (data not shown).

Immediately prior to aconitase and fumarase activity measurements, mitochondrial fractions were resuspended in 0.5 mL buffer containing 50 mmol/L Tris–HCl (pH 7.4) containing 0.6 mmol/L MnCl2 and sonicated for 2 s. Aconitase activity was immediately measured spectrophotometrically by monitoring the formation of cis-aconitate from isocitrate at 240 nm in 50 mmol/L Tris–HCl (pH 7.4) containing 0.6 mmol/L MnCl2 and 20 mmol/L isocitrate at 25°C. One unit was defined as the amount of enzyme necessary to produce 1 μmol cis-aconitate per minute. Fumarase activity was measured by monitoring the increase in absorbance at 240 nm at 25°C in a 1 mL reaction mixture containing 30 mmol/L potassium phosphate (pH 7.4) 0.1 mmol/L l-malate. One unit was defined as the enzyme necessary to produce 1 μmol fumarate per minute. Protein concentrations were measured using Coomassie Plus reagents (Pierce Biotechnology, Rockford, IL, USA). The desired protein concentrations range in samples for the aconitase assay is 0.5–2 mg/mL.

Stereotactical implantation of lithium phthalocyanine into the brain

Sprague–Dawley rats weighing 200–250 g (2–3-month-old) were used for the experiments. Approximately 1 week to 10 days prior to EPR measurements rats were anesthetized with 3.5–4.0% isoflurane in 26% oxygen. Oxygen-sensitive lithium phthalocyanine (LiPc) crystals synthesized in the EPR Center (Dartmouth Medical School, Hanover, NH, USA) were implanted according to procedures described previously (Hou et al. 2005). Briefly, LiPc crystals (50–80 μg) were placed via 23-gauge needles into the brain using stereotactic techniques using the following coordinates: hippocampus (on the right side) from bregma AP = −3.8 mm, ML = 3.0 mm, and DV = 3.8 mm and cortex (on the left side): AP = −3.8 mm, ML = 1.5 mm and DV = 1.2 mm.

Cerebral pO2 measurements

Oxygen levels in the hippocampus and cortex of rats were analyzed by EPR at different time points, including baseline (before kainate injection), 6, 24, 48, and 72 h after kainate injection. The rats were anesthetized with isoflurane (1.2–1.5%) in 26% oxygen and placed in the magnet. EPR oximetry was performed using a low frequency L-band (1.2 GHz) spectrometer that was developed specifically for in vivo studies. Spectra were detected using a surface loop resonator, which was positioned over the region of interest. The settings for the spectrometer were: non-saturating microwave power; magnetic field center, 425 Gauss; scan range, 0.5–5.0 Gauss; modulation frequency, 27 kHz. Modulation amplitude was set at less than one-third of the EPR line width. Scan time was 10 s, and 10 scans were performed for each site to increase the signal to noise ratio. Heart rate and the blood oxygen saturation were continuously and non-invasively monitored by a pulse oximeter (Nonin Medical, Plymouth, MN, USA) during all EPR measurements. Body temperature was maintained within the normal range (37–38°C) using a heated pad, and the core temperature of the rat was monitored by a rectal probe.

Histopathological examination

After the last EPR measurements, rats were killed with CO2. The brains were removed immediately and stored in a 10% formalin solution until histological evaluation was performed. LiPc crystal implantation sites were verified by histological staining.

Statistical analysis

Statistical differences were analyzed by one-way anova with post hoc multiple comparison test (Tukey test). p-Values < 0.05 were considered statistically significant.


Kainate-induced SE increases the formation of IsoFs and F2-IsoPs with overlapping but distinct time courses

We previously showed that kainate-induced SE results in a subregion and time-dependent formation of F2-IsoP in the rat hippocampus (Patel et al. 2001). As mentioned, it has been shown that the formation of F2-IsoPs and IsoFs are differentially modulated by oxygen tension (Fessel et al. 2002). Therefore, in this study, we determined the time course of SE-induced IsoF formation and assessed its temporal relationship with F2-IsoP formation. SE produced a significant increase in both F2-IsoP and IsoF with overlapping but distinct time courses. In the CA1 and CA3 regions, F2-IsoP formation was significantly increased only at the 16 h time point, whereas IsoF formation was significantly increased both at the 16 and 48 h time points (Fig. 1a). In the DG, F2-IsoP formation was significantly increased by 8 and 16 h whereas IsoF formation was significantly increased 16 and 48 h returning to near-baseline values 7 days following kainate injection (Fig. 1a). Cerebellar F2-IsoP and IsoF levels remained unchanged at all time points following kainate injection (Fig. 1a).

Figure 1.

 (a) Rats were injected with kainic acid (12 mg/kg, s.c.) and tissue-esterified F2-IsoP and IsoF levels were measured in hippocampal CA1, CA3, DG regions, and cerebellum by gas chromatography–mass spectrometry as described in Methods. Each point represents mean ± SEM (n = 6–10 rats/group, *p < 0.05 one-way anova). (b) Data is expressed as ratio of IsoF : F2-IsoP.

Previous assessment of IsoF and F2-IsoP concentrations in different tissues suggests a strong correlation between the ratio of IsoF : F2-IsoP and oxygenation of the tissue suggesting that changes in the ratio of IsoF : F2-IsoP may serve as a dynamic putative index of tissue oxygenation (Fessel et al. 2002; Fessel and Jackson Roberts 2005). For example, highly oxygenated tissues such as the brain and kidney have significantly higher ratios of IsoF : F2-IsoP compared with the liver, a poorly oxygenated organ. Analysis of the ratios of IsoF : F2-IsoP following SE, revealed that the ratio IsoF : F2-IsoP in all hippocampal subregions was significantly increased at the 48 h time point (Fig. 1b), whereas ratio of IsoF : F2-IsoP was unchanged in the cerebellum (Fig. 1b).

Seizure-induced formation of IsoFs, but not F2-IsoPs correlates with the occurrence of mitochondrial oxidative stress

The data so far suggest that SE-induced F2-IsoP formation precedes but overlaps with the formation of IsoFs. As (i) IsoF formation is favored by higher tissue oxygen tension and (ii) mitochondrial dysfunction can decrease oxygen utilization and thereby increase local oxygen tension, we hypothesized that the formation of IsoF, but not F2-IsoP would more closely correlate with mitochondrial oxidative stress. To test this idea, we measured the activities of hippocampal mitochondrial aconitase and fumarase at different time points following SE and correlated its inactivation, an index of mitochondrial reactive oxygen species (ROS) production, with the formation of F2-IsoP and IsoF formation in the CA3 region. The correlation revealed that in comparison with F2-IsoP formation, SE-induced formation of IsoF more closely correlated with mitochondrial aconitase inactivation, a previously validated index of mitochondrial oxidative stress in this model (Liang et al. 2000) (Fig. 2). The activity of fumarase, an oxidatively stable control enzyme was unchanged at times during which aconitase activity was decreased (data not shown). The time courses of SE-induced formation of IsoFs but not F2-IsoPs also closely correlated with another previously reported index of mitochondrial oxidative stress, reduced glutathione levels in mitochondrial fractions (Liang and Patel 2006).

Figure 2.

 Correlation of the time course of SE-induced hippocampal CA3 subregion IsoF and F2-IsoP formation with the activity of mitochondrial aconitase. Data points obtained 0–48 h in Fig. 1 are plotted as percent of control (a and c). Mitochondrial aconitase activity (b) was measured as described in Methods and plotted as percent inhibition of control values (n = 6 rats per time point, p < 0.01, one-way anova).

Seizure-induced formation of F2-IsoP and IsoFs correlates with changes in hippocampal pO2

To determine if changes in hippocampal oxygen tension influenced the formation of F2-IsoPs and IsoFs following SE, we determined brain oxygen tension by an EPR oximetry method (Swartz and Clarkson 1998). Measurement of oxygen levels in the hippocampus and cortex of the rats 6, 24, 48, and 72 h after kainate injection revealed a significant decrease in hippocampal (approximately 42% from baseline values), but not cortical oxygen levels at the 6 h time point (Fig. 3a and b). This suggests that SE produces a period of transient hypoxia in the hippocampus. Correlation of SE-induced changes in oxygen levels with F2-IsoP and IsoF formation revealed that F2-IsoP formation coincided with the early hypoxic phase; whereas IsoF formation occurred during the phase when the tissue oxygen levels return to normal values (Fig. 4a and b).

Figure 3.

 The levels of oxygen in the hippocampus (a) and the cortex (b) of rats at different time points after kainic acid injection (12 mg/kg, s.c.). Points represent mean ± SEM. Each point is the average of 3–4 rats; *p < 0.05 one-way anova.

Figure 4.

 Correlation of IsoFs (a) and F2-IsoP (b) levels with changes in CA3 area with pO2 levels in the hippocampus. Left y-axis is percent change of IsoF or F2-IsoP. Right y-axis is percent change in pO2; x-axis is time after kainic acid (0 = control).


Three major findings arise from this study. First, SE results in the formation of lipid peroxidation end products, IsoFs and F2-IsoPs, with distinct, but overlapping time courses. Secondly, SE-induced IsoF formation correlates closely with the occurrence of mitochondrial aconitase inactivation, an index of mitochondrial oxidative stress. Finally, SE transiently decreases hippocampal oxygen tension and SE-induced formation of F2-IsoP is predominant during the hypoxic phase, whereas the formation of IsoF occurs when the tissue oxygen levels return to normal. These data suggest that seizure-induced changes in tissue oxygen levels and mitochondrial dysfunction may differentially influence the formation of F2-IsoPs and IsoFs and the combined measurement of F2-IsoPs and IsoFs provides a more reliable assessment of seizure-induced oxidative stress and changes in tissue oxygenation.

In this study we show that kainate-induced seizures result in a time-dependent increase in IsoF formation in hippocampal subregions. In a previous study we have shown seizure-induced formation of F2-IsoPs (Patel et al. 2001). Here we measured the time course of IsoFs and compared it with the time course of F2-IsoP formation. The time course and hippocampal subregion dependence of seizure-induced F2-IsoP formation was similar to that observed in our previous study. Interestingly, in the present study the formation F2-IsoP and IsoF showed overlapping but distinct time courses. The magnitude of SE-induced IsoF (and F2-IsoP) formation was greater in the DG compared to CA1 and CA3; whereas no changes were observed in the cerebellum. This is consistent with seizure activity being the stimulus for the formation of both products.

While the F2-IsoPs are a valuable validated biomarker for assessing oxidative stress in vivo, recent studies suggest that their formation is modulated by oxygen tension in that as oxygen tension increases the formation of IsoFs becomes favored where as the formation of F2-IsoPs becomes disfavored and vice versa when oxygen tension decreases (Fessel et al. 2002; Roberts and Fessel 2004). Consistent with this, increasing pO2 from 1% to 21% during oxidation of arachidonic acid in vitro markedly increases the formation of IsoFs but the effect on the formation of F2-IsoPs was minimal (Fessel et al. 2002). Moreover, IsoFs but not F2-IsoP levels were significantly higher in the lungs of mice exposed to 100% pO2 in comparison to 21% pO2 (Roberts et al. 2005). It has been suggested that the formation of IsoF may be favored in pathophysiological states associated with oxidative stress and mitochondrial dysfunction. Cellular oxygen concentrations measured by LiPc EPR oximetry increase or decrease under conditions that decrease or increase oxygen consumption, respectively (Pandian et al. 2003). Furthermore, increased formation of IsoFs, but not F2-IsoPs was found in the substantia nigra from patients with Parkinson’s disease (Fessel et al. 2003), a disease in which mitochondrial dysfunction is well established (Schapira et al. 1998; Beal 2005). Consistent with this, we observed that seizure-induced IsoF formation correlated closely with inactivation of mitochondrial aconitase, an index of mitochondrial oxidative stress, whereas F2-IsoP formation did not.

The observation that the formation of F2-IsoPs and IsoFs shows overlapping, yet distinct time courses following seizures suggests that this may be as a result of local changes in hippocampal pO2 levels. Measurement of brain pO2 levels following kainate administration confirmed transient, but significant hypoxia occurs following SE in the hippocampus but not cortex. Seizure-induced formation of F2-IsoPs began during the hypoxic phase and returned to baseline as pO2 levels recovered. This is consistent with previous studies demonstrating that F2-IsoP formation occurs under low pO2 and is disfavored at higher tissue oxygen levels. The time course of seizure-induced formation of IsoF formation correlated well with the return to normal pO2 levels in the hippocampus. This is consistent with the dependence of IsoF formation on tissue pO2 levels.

To our knowledge this is the first study showing specific decreases in hippocampal pO2 in the kainate SE model. Although oxygen sufficiency during seizures has been controversial (Meldrum et al. 1973; Kreisman et al. 1984), several studies have suggested that seizure activity is followed by a decrease in pO2 indicative of a hypoxic phase (Kreisman et al. 1983, 1991; Hoshi and Tamura 1993). An episode of ‘ictal ischemia’ has been observed following neocortical seizures in rats induced by 4-aminopyridine (Bahar et al. 2006). Moreover, cortical free fatty acids and diacylglycerol also followed the pattern of early increase followed by late decrease following seizure activity (Visioli et al. 1993). Previous studies have validated the measurement of brain pO2 levels using LiPc oxygen biosensors (O’Hara et al. 2005; Dinguizli et al. 2006). In vivo EPR oximetry is a powerful approach to monitor tissue oxygenation in animals and human subjects. The assessment of the pO2 is based on the variation of the EPR line widths of spectra recorded from implanted particulate paramagnetic probes which are highly dependent on changes in pO2 (Dunn and Swartz 2003). The changes observed in pO2 levels using the LiPc oximetry validate the use of this methodology (Liu et al. 2004a; Hou et al. 2005) and confirm earlier work in other seizure models that demonstrated a decrease in pO2 levels during early time period after SE (Kreisman et al. 1983, 1991; Hoshi and Tamura 1993). As we were unable to measure pO2 during the acute phase of SE (< 6 h) it is difficult to ascertain when the changes in pO2 began. It may be speculated that the dip in pO2 accompanied the SE and pO2 returned to normal as seizure activity ceased. The specificity of pO2 changes in the hippocampus and not the cortex correlates with the specificity of kainate-mediated seizure activity in the hippocampal region and further suggests that seizure activity may be the major trigger of the hypoxia.

Mechanisms that underlie the development of hypoxia during SE remain to be determined; however, seizure-induced expression of vascular factors such as vascular endothelial growth factor (Newton et al. 2003) is likely a result of the seizure-induced hypoxia. An important role of SE-induced hypoxia and subsequent reoxygenation may be the generation of ROS. We have conducted extensive studies to determine the time course and cellular sites of SE-induced ROS production (Liang et al. 2000; Patel et al. 2001; Patel and Li 2003; Liang and Patel 2004). These studies show that the earliest increase in SE-induced ROS production begins approximately 8 h after the injection of kainate and is maximal 16–24 h thereafter which inversely coincides with the hypoxic phase observed here. Thus the production of ROS and oxidation of susceptible targets is low during the hypoxia phase and ROS formation may be occurring as a result of reoxygenation of the tissue. Interestingly, the magnitude of hippocampal hypoxia observed here was similar to that observed using LiPc oximetry in the ischemic penumbra in the rat middle cerebral artery occlusion model (Liu et al. 2004b).

In summary, we show that SE results in the formation of IsoFs in seizure-prone hippocampus, in a manner that correlates with the occurrence of mitochondrial oxidative stress and reoxgyenation following a period of transient hypoxia. In contrast to the enzymatic formation of arachidonic acid metabolites by SE, we show here SE-induced non-enzymatic formation of arachidonic acid-derived products with potentially important biological activities (Morrow et al. 1994). The formation of IsoFs and F2-IsoPs not only contribute to the many inflammatory changes occurring following SE (Bazan et al. 1986; Patel et al. 2005; Vezzani and Granata 2005) but may have profound consequences on SE-induced pathophysiology.


This study was supported by NIH grant NS39587 (MP), NIH grant GM42056 (LJR), NIH (NIBIB) grant P01 EB002180 (HMS), and the facilities of the EPR Center for the Study of Viable Systems supported by NIH (NIBIB) grant P41 EB002032 (HMS).