Authors Busceti, Biagioni, and Aronica contributed equally to this work.
Address correspondence and reprint requests to Ferdinando Nicoletti, Department of Human Physiology and Pharmacology, P.le Aldo Moro 5, 00185 Roma, Italy. E-mail: firstname.lastname@example.org
Summary: Inhibition of the Wnt pathway by the secreted glycoprotein, Dickkopf-1 (Dkk-1) has been related to processes of excitotoxic and ischemic neuronal death. We now report that Dkk-1 is induced in neurons of the rat olfactory cortex and hippocampus degenerating in response to seizures produced by systemic injection of kainate (12 mg/kg, i.p.). There was a tight correlation between Dkk-1 expression and neuronal death in both regions, as shown by the different expression profiles in animals classified as “high” and “low” responders to kainate. For example, no induction of Dkk-1 was detected in the hippocampus of low responder rats, in which seizures did not cause neuronal loss. Induction of Dkk-1 always anticipated neuronal death and was associated with a reduction in nuclear levels of β-catenin, which reflects an ongoing inhibition of the canonical Wnt pathway. Intracerebroventricular injections of Dkk-1 antisense oligonucleotides (12 nmol/2 μL) substantially reduced kainate-induced neuronal damage, as did a pretreatment with lithium ions (1 mEq/kg, i.p.), which rescue the Wnt pathway by acting downstream of the Dkk-1 blockade. Taken collectively, these data suggest that an early inhibition of the Wnt pathway by Dkk-1 contributes to neuronal damage associated with temporal lobe epilepsy. We also examined Dkk-1 expression in the hippocampus of epileptic patients and their controls. A strong Dkk-1 immunolabeling was found in six bioptic samples and in one autoptic sample from patients with mesial temporal lobe epilepsy associated with hippocampal sclerosis. Dkk-1 expression was undetectable or very low in autoptic samples from nonepileptic patients or in bioptic samples from patients with complex partial seizures without neuronal loss and/or reactive gliosis in the hippocampus. Our data raise the attractive possibility that drugs able to rescue the canonical Wnt pathway, such as Dkk-1 antagonists or inhibitors of glycogen synthase kinase-3β, reduce the development of hippocampal sclerosis in patients with temporal lobe epilepsy.
Hippocampal sclerosis is typically associated with mesial temporal lobe epilepsy, and is characterized by neuronal loss and reactive gliosis, most prominently in the CA1 field of the hippocampus, followed by the hilus, CA4, and CA3 fields. Neurons in the dentate granular cell layer and in the CA2 field are relatively unaffected (Honavar and Meldrum, 1997). Although this association has been known for a long time and complicates the clinical outcome of temporal lobe epilepsy, the molecular mechanisms by which seizures cause hippocampal damage are only partially identified. Seizure-induced neuronal damage involves an excitotoxic component (Fuller and Olney, 1981), suggesting that ionotropic glutamate receptor antagonists are potential neuroprotective agents in epilepsy. These drugs, however, are not suitable for long-term treatment in humans because they induce serious adverse effects resulting from the inhibition of fast excitatory synaptic transmission, such as sedation, ataxia, and impairment of learning and memory (Loscher, 1998; Bruno et al., 2001). The identification of specific processes that lie downstream of glutamate receptors along the death pathway may provide new targets for safer neuroprotective agents. Recent evidence indicates that excitotoxic neuronal death in culture and hypoxic/ischemic neuronal damage in vivo is causally related to the expression of the secreted glycoprotein Dickkopf-1 (Dkk-1) (Cappuccio et al., 2005). Dkk-1 acts as a selective antagonist of the canonical Wnt pathway by interacting with the low-density lipoprotein receptor-related proteins (LRP) 5 or 6, which are also receptors for Wnt glycoproteins. In the canonical Wnt pathway, Wnt glycoproteins bind to the 7-TM receptor, Frizzled, and to LRP5/6, thus triggering a cascade of intracellular events that inhibit the activity of glycogen synthase kinase-3β (GSK-3β) and prevent β-catenin ubiquitination and degradation (Willert and Nusse, 1998). As a consequence β-catenin migrates to the cell nucleus where it promotes the transcription of genes that are important for neuronal homeostasis and survival. The canonical Wnt pathway is negatively modulated by the secreted protein Dkk-1, which binds to LRPs, thus preventing their interaction with Wnts (Zorn, 2001; Mao et al., 2002; Grotewold and Ruther, 2002a,b).
Dkk-1 has a prominent role during development, and is expressed at very low levels in the adult brain (Diep et al., 2004). Its expression increases in response to insults that cause DNA damage, such as excitotoxic insults, because Dkk-1 is transcriptionally regulated by p53 (Wang et al., 2000). Hence, induction of Dkk-1 might represent a component of the sequence of events leading to neuronal death (Caricasole et al., 2003). We recently found an induction of Dkk-1 in cultured neurons challenged with β-amyloid peptide as well as in degenerating neurons of the Alzheimer's brain (Caricasole et al., 2004) and in neurons subjected to excitotoxic or ischemic insults and is required for the development of ischemic neuronal damage (Cappuccio et al., 2005).
Here, we report that Dkk-1 is early induced in cortical and hippocampal neurons vulnerable to kainate-induced seizures, and its expression is causally related to neuronal death. In addition, Dkk-1 is specifically found in degenerating neurons of patients affected by mesial temporal lobe epilepsy associated with hippocampal sclerosis.
Adult male Sprague-Dawley rats (Harlan, Udine, Italy), weighing 150–170 g, were maintained in a temperature- and light-controlled environment with a 14/10 h light/dark cycle and were treated in accordance with the principles and procedures of the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. Animals were injected i.p. with 12 mg/kg of kainate (Tocris Coockson, Bristol, U.K.; dissolved in PBS, pH 7.4). Control rats were injected with PBS alone. In additional groups of animals, kainate was injected after a pretreatment with lithium chloride or with Dkk-1 antisense oligonucleotides. Lithium chloride was injected i.p. every 12 h at the dose of 1 mEq/kg for the 7 days preceding the injection of kainate, and then for the following 7 days. Dkk-1 antisenses or the corresponding sense oligonucleotides were injected i.c.v. (12 nmol/2 μL) 24 h and 1 h before the injection of kainate and then once a day for the following 3 days through a guide cannula implanted under thiopental anesthesia (coordinates: AP –0.8 mm; ML 1.4 mm according to the Paxinox and Watson Atlas, 1982). The following end-capped phosphorothioate rat Dkk-1 antisense oligonucleotide was used: 5′-CGTCGGAGGGAGGCGTGC-3′; control rats were injected with the corresponding sense oligonucleotide: 5′-GCTCGCCTCCCTCCGACG-3′ (Invitrogen, Milano, Italy). Oligonucleotides were dissolved in sterile saline.
Behavioral assessment of motor seizures
Motor seizures were observed for 4 h following kainate injection, and scored according to Racine (1972), as follows: 0 = absence of seizures; 1 = staring spells, immobilization, and hypoactivity; 2 = paroxysmal wet dog shake and head nodding; 3 = motor seizures associated with masticatory movements and tail arching; 4 = rearing with forelimb jerks and salivation; 5 = generalized convulsions with loss of postural control and intense myoclonic jerks lasting at least 1 h; and 6 =“full status epilepticus” and death. Animals with a score of ≤3 were arbitrarily classified as “low” responders; animals with a score of ≥4 were considered as “high” responders.
We carried out EEG recording (a) in five animals treated with kainate (retrospectively classified as low or high responders) for the characterization of our model; and (b) in four animals pretreated with saline for 5 days and then challenged with kainate or in six animals pretreated with lithium chloride (1 mEq/kg/12 h) for 5 days and then challenged with kainate. Rats were anesthetized with chloral-hydrate (400 mg/kg, i.p.) and placed in a Kopf stereotaxic apparatus. For each rat, four recording stainless steel screws epidural electrodes were implanted. Two electrodes were placed bilaterally above the frontal cortex (ML =+ 2 mm from midline, AP =+ 4 mm from bregma), and two electrodes were placed bilaterally above the posterior cortex (ML =+3 mm from midline, AP =−5.5 from bregma). For each rat, two additional similar electrodes were implanted over the cerebellum and were used as reference and ground electrode, respectively. Electrodes were kept in place by dental acrylic cement. After surgery, rats were allowed to recover for 24 h in their home cages with free access to food and water. EEGs were recorded at least 72 h after surgery. On the day of recording the female electrode microconnectors were connected via shielded cables to a dedicated digital EEG apparatus (BE-Light, EBNeuro, Florence, Italy). Before each recording session, electrodes impedance was measured and only signals recorded from electrodes with impedance <3 kω were used for analysis. EEG signals collected were notch filtered, low- and high-pass filtered (64 Hz and 0.5Hz, respectively), and stored with a 128 Hz sampling rate on a PC equipped with the software Galileo-NT (EBNeuro), for off-line analysis. On the day of each recording session, rats were placed in transparent Plexiglas cylinder cages at least 30 min before starting EEG recording. Rats were recorded for 30 min/day in baseline conditions, and during chronic lithium administration. On day of seizure induction, EEG recording was started 10 min before kainate administration and continued up to 12 h later. Seizure episodes and their behavioral score (see above) were real-time annotated on the PC-imported EEG trace.
EEG power-spectrum analysis
Power spectral analysis was carried out using two referential EEG traces: left anterior versus reference (abbreviated as “FP1–RF” in figures), and right posterior versus reference (abbreviated as “O2–RF” in figures). The spectral analysis of the EEG was performed by computing the fast Fourier transformation (Cooley and Tukey, 1965), via the spectral analysis software of GAL-NT (EBNeuro). In particular, the mean EEG power spectrum for the range of frequencies between 0.1 and 64 Hz was calculated from 64 s-long segments, by averaging the power spectrum of its 8-s long consecutive epochs. For recordings collected during baseline and chronic lithium administration (see above), the analyzed segments were randomly selected from EEG segments recorded during waking immobility to avoid potential movement artifacts. For EEG traces collected after kainate injection, the analyzed segments were selected from EEG segments recorded during seizures. For each rat, after visual inspection of the EEG traces collected during SE, we collected 64-s-long segments representative of the main electrographic pattern activity.
For each animal, data from three randomly selected 64 s-segments were used. Each power spectrum obtained from the 64 s-long segments was subdivided in groups of frequencies ranging 3 Hz, starting from 0 Hz (1st = 0–3 Hz, 2nd = 3–6 Hz, 3rd = 6–9 Hz, and so on, up to 20 Hz) and the absolute and relative power was calculated for each range of frequencies and expressed as spectral power ± standard deviation, or mean percentage relative power ± standard deviation, respectively. For each frequency range, the relative powers from different experimental groups were compared with analysis of variance (ANOVA). Null hypothesis was discarded when p < 0.05.
Assessment of Dkk-1 expression and β-catenin levels
Dkk-1 expression was examined in the olfactory cortex and hippocampus on rats killed from 6 h to 7 days following kainate injection. Rats treated with Dkk-1 antisense or sense oligonucleotides were killed at 3 days after kainate injection. Dkk-1 expression was assessed by immunohistochemistry or Western blot analysis; β-catenin levels were assessed by Western blot analysis.
Dissected brains were fixed in Carnoi, embedded in paraffin, and sectioned at 10 μm. Subsequently, deparaffinized sections were immersed in 0.3% H2O2 in methanol for 30 min to quench endogenous peroxide activity, treated in 10 mM, pH 6.0, citrate buffer, and heated by microwave for 10 min for antigen retrieval. The slides were allowed to cool for 20 min in the same solution at room temperature and then washed in PBS. Normal serum and antibodies were dissolved in PBS containing 10% normal horse serum. Sections were preincubated for 30 min with normal horse serum (Dako, Glostrup, Denmark), treated with primary Dkk-1 antibodies obtained either from Santa Cruz Biotechnology (SC-14949; goat polyclonal; 1:10) or from R&D Systems (DBA, Milan, Italy; AF1765; goat polyclonal; 1:10), and then for 1 h with secondary biotinylated antigoat antibodies (1:200; Vector Laboratories, Burlingame, CA, U.S.A.). For detection, 3,3-diaminobenzidine tetrachloride was used (ABC Elite kit; Vector Laboratories). Control staining was performed without the primary antibodies.
Western blot analysis
Western blot analysis of Dkk-1 and β-catenin expression was performed in protein lysates obtained from the entire hippocampus of rats subjected to kainate injection. Dkk-1 protein (35 kDa) was detected with a goat polyclonal antibody (R&D Systems; 1:50). In some experiments, either Dkk-1 or β-catenin were detected in nuclear and cytosolic fractions isolated by differential centrifugation using a commercially available kit (Pierce, Rockford, IL, U.S.A.) (Calderone et al., 2003). The β-catenin detection was performed by using mouse β-catenin antibody (1:1,000; BD Transduction Laboratories, Beverly, MA, U.S.A.). The purity of the fractions was verified with an antibody against the nuclear protein poly(ADP-ribose) polymerase (Santa Cruz Biotechnology, Santa Cruz, CA, U.S.A.). Immunoreactive bands were revealed by ECL (Amersham Biosciences, Milano, Italy).
Assessment of neuronal death
Histology and quantitative analysis
Dissected brains were processed as described earlier. Sections (10 μm) were processed for staining with thionin (Nissl staining). For the assessment of neuronal death, the number of surviving neurons in the olfactory cortex and the hippocampus was counted under a light microscope at 20× magnification. Neurons with a rounded shape similar to that commonly observed in sections from control animals were considered to be viable. We used a nonstereological method for the assessment of neuronal density (neurons per mm3 of tissue, Nv) using the following formula: Nv =NA/(t+D), where NA is the number of neurons per mm2 of tissue, t is the section thickness, and D is the neuron diameter (Abercrombie, 1946). For each control (n = 3) and kainate-treated animal (n = 6), 20 sections were sampled at the same regular intervals.
Expression of Dkk-1 in the hippocampus of human patients with temporal lobe epilepsy associated with hippocampal sclerosis
The cases included in this study were obtained from the files of the Departments of Neuropathology of the Academical Medical Center (AMC, University of Amsterdam), the VU University Medical Center (VUMC) in Amsterdam. Ten patients (six with and four without hippocampal sclerosis) underwent resection of the hippocampus for medically intractable epilepsy. Autoptic hippocampal samples were available from one additional patient with hippocampal sclerosis who died during a status epilepticus. In all patients with hippocampal sclerosis, the lesion was localized by brain MRI. All cases were reviewed independently by two neuropathologists and the diagnosis was confirmed according to the criteria summarized in the recently commissioned International League Against Epilepsy report (Wieser, 2004), including the detection of neuronal cell loss and gliosis (CA1 and endfolium) and the detection of synaptic reorganization. Control hippocampal tissue was obtained at autopsy from six patients without history of seizures or other neurological diseases. All autopsies were performed within 12 h after death. Informed consent was obtained for the use of brain tissue and for access to medical records for research purposes. Tissue was obtained and used in a manner compliant with the Declaration of Helsinki.
Formalin fixed, paraffin-embedded tissue was sectioned at 6 μm and mounted on organosilane-coated slides (Sigma, St. Louis, MO, U.S.A.). Representative sections of all specimens were processed for hematoxylin-eosin and for immunohistochemical detection of Dkk-1 performed by using the following antibodies: anti-Dkk-1 obtained from Santa Cruz Biotechnology (goat polyclonal, raised against an internal region of human Dkk-1; 1:100, Santa Cruz, CA, U.S.A.) and from Abcam (goat polyclonal; raised against the C terminus of Dkk-1; 1:100, Abcam, Milano, Italy). Immunohistochemistry was carried out as previously described using avidin–biotin peroxidase method (Vector Elite, Vector Laboratories, Burlingame, CA, U.S.A.) and 3,3′-diaminobenzidine as chromogen (Aronica et al., 2003; Caricasole et al., 2004). Sections incubated without the primary antibody, with preimmune sera or with the antibody preincubated with the antigenic peptide were essentially blank.
Kainate-induced seizures and correlation to neuronal damage
Systemic injection of kainate (12 mg/kg, i.p.) induced limbic motor seizures evolving into generalized tonic–clonic seizures and status epilepticus in about 70% of the animals. The seizure severity score in this subgroup of rats arbitrarily classified as high responders was ≥4 (average score = 4.74 ± 0.71; means ± S.E.M; n = 27). In the remaining 30% of kainate-injected rats, the seizure severity score was ≤3 (average score = 2.15 ± 0.96; means ± S.E.M; n = 13). These animals were considered as low responders. Typical EEG traces with power spectrum analysis in representative high and low responder rats (seizure severity score = 1 and 5, respectively) are shown in Fig. 1. Neuronal death was assessed by nonstereological counts of Nissl-stained neurons in the olfactory cortex and hippocampus, two brain regions that are particularly sensitive to seizure-induced neuronal damage. In the olfactory cortex of high responder rats, neuronal death was already substantial as early as 12 h after kainate injection, and further increased between 24 h and 7 days (Fig. 2). In the olfactory cortex of low responder rats, neuronal death developed with a longer latency and became visible after 24 h (Fig. 3). Only high responder rats showed neuronal damage in the hippocampus. The number of CA1 neurons was apparently reduced by about 55–60% at 3 days and by about 75% at 7 days after kainate injection. No death was detectable before 3 days (Fig. 4). Neuronal death was occasionally detected in the CA4 region (not shown).
Induction of Dkk-1 precedes neuronal death in kainate-injected rats
Dkk-1 expression was assessed by immunohistochemistry in sections adjacent to those used for Nissl staining using the goat anti-Dkk-1 Santa Cruz antibodies after pretreating the sections with citrate buffer, as reported in the Methods section. In cortical neurons of high responder rats, expression of Dkk-1 substantial increased at 6 h, when cell damage was not yet detectable. Expression remained high in surviving neurons at later times (Fig. 2). At cellular level, Dkk-1 immunostaining was mainly detectable in the cytoplasmic and in the perinuclear region, although punctuate nuclear staining was occasionally observed (Fig. 2). An early induction of Dkk-1 was also observed in the olfactory cortex of low responder rats at 12 h, a time that precedes neuronal death (Fig. 3). In the hippocampus of high responder rats, Dkk-1 was induced at 12–24 h, again preceding neuronal death (Fig. 4). No induction of Dkk-1 was found in the hippocampus of low responder rats (not shown). The cytoplasmic localization of Dkk-1 staining found with the Santa Cruz antibody was confirmed using the R&D Systems anti-Dkk-1 antibody both in the olfactory cortex and hippocampus of high responder rats (Fig. 5). The early induction of Dkk-1 in the hippocampus of high responder rats was confirmed by Western blot analysis (Fig. 6A). Immunoblots showed a single band at 35 kDa, corresponding to the deduced molecular size of Dkk-1 (Grotewold and Ruther, 2002b). The intensity of this band increased substantially at 12 h following kainate injection and progressively decreased afterwards, perhaps as a result of the ongoing neuronal death (Fig. 6A). Dkk-1 was selectively expressed in the cytoplasmic fraction isolated from hippocampal extracts, and was not found in the nuclear fraction, in agreement with immunohistochemical data (Fig. 6B). No changes in Dkk-1 expression were detected by immunoblotting in the hippocampus of low responder rats (Fig. 6A).
We extended the analysis to β-catenin levels, which are regulated by the canonical Wnt pathway (Cadigan and Nusse, 1997). As activation of the Wnt pathway increases the stability and nuclear translocation of β-catenin, we expected to detect an early reduction of nuclear β-catenin following kainate injection. Accordingly, we found a reduction of β-catenin levels in the hippocampal nuclear fraction of high responder rats, which was already substantial at 12 h (Fig. 6C), and thus paralleled the increase in Dkk-1 expression. No changes in β-catenin levels were found in the cytoplasmic fraction (Fig. 6C).
Lithium treatment and Dkk-1 knockdown protect against seizure-induced neuronal death in kainate-injected rats
To examine whether induction of Dkk-1 contributed to neuronal damage associated with kainate-induced seizures, we pretreated animals with either lithium ions or Dkk-1 antisense oligonucleotides. Lithium ions are known to rescue the canonical Wnt pathway by inhibiting GSK-3β that is, by acting downstream of the Dkk-1 blockade (Klein and Melton, 1996). Lithium was injected i.p. as LiCl (1 mEq/kg, twice daily), for the 7 days preceding kainate injection, and then for the following 7 days. Power spectrum analysis showed that lithium treatment did not change EEG activity during the pretreatment period (Fig. 7A) as well as following kainate injection (Fig. 7B). In addition, lithium treatment did no change the percentage of rats considered as high and low responders to kainate (high responders = 68% and 70% in lithium-treated and control rats, respectively). Lithium treatment substantially protected vulnerable neurons of the olfactory cortex (Fig. 8A) and CA1 region (Fig. 8B) against seizure-induced damage examined at 7 days following kainate injection. In additional groups of animals, end-capped Dkk-1 antisense oligonucleotides or the corresponding sense oligonucleotides were injected i.c.v. (12 nmol/2 μL) 24 h and 1 h prior to kainate injection, and then once a day for the following 3 days. Antisense treatment did not affect kainate-induced seizures, but prevented both Dkk-1 expression and neuronal death in the olfactory cortex and hippocampus at 3 days. No changes were induced by treatment with sense oligonucleotides (Fig. 9A,B). The protective effects of lithium or Dkk-1 antisenses against kainate-induced neuronal death were substantial although the value of these data might be limited by the use of a nonstereological method of neuronal counting.
Dkk-1 expression in the hippocampus of patients with temporal lobe epilepsy
We examined Dkk-1 expression in the following human samples: (a) bioptic hippocampal tissue from six patients with mesial temporal lobe epilepsy associated with hippocampal sclerosis; (b) autoptic hippocampal tissue from one patient with temporal lobe epilepsy associated with hippocampal sclerosis died during status epilepticus; (c) bioptic hippocampal tissue from four epileptic patients with complex partial seizures without histological evidence of neuronal loss or reactive gliosis; and (d) autoptic hippocampal tissue from 6 nonepileptic controls. The clinical characteristics of all subjects are summarized in Table 1. No expression of Dkk-1 was detected in the CA1, dentate gyrus (DG), or CA3 region of autoptic control tissue (Fig. 10A, G, I). All samples (bioptic or autoptic) with histological evidence of hippocampal sclerosis showed a robust expression of Dkk-1 in the hippocampus (Fig. 10B, F, H, N). Expression was detectable in the cytoplasmic region of surviving neurons and in the extracellular space, in agreement with the role of Dkk-1 as a secreted protein (Kawano and Kypta, 2003). Expression of Dkk-1 was very low or absent in three of the four bioptic samples from epileptic patients without histological signs of hippocampal sclerosis (examples are shown in Fig. 10C, D, M). The hippocampus of the remaining patient showed a moderate expression of Dkk-1, which, however, was lower than the expression found in samples with hippocampal sclerosis (compare Fig. 10E with Fig. 10F).
Table 1. Clinical characteristics of epilepic patients
Systemic or intracerebral injections of kainate cause epileptiform seizures that propagate from the CA3 region of the hippocampus to other limbic regions and are associated with a pattern of neuronal loss reminiscent of that seen in patients with temporal lobe epilepsy (reviewed by Nadler, 1981; Ben-Ari and Cossart, 2000). The regional profile of neuronal loss strictly depends on the route of administration of kainate, and other variables including the strain and age of animals (reviewed by Ben-Ari and Cossart, 2000). When kainate is injected i.c.v. or is intracerebrally infused even in regions that are distant from the hippocampus, brain damage mainly involves the hippocampal formation, and, in particular, pyramidal neurons of the CA3 region (Nadler and Cuthbertson, 1980; Nadler, 1981; Ben-Ari, 1985; Ben-Ari and Cossart, 2000). When kainate is injected systemically, the olfactory cortex and amygdala are more vulnerable to neuronal damage than the hippocampal formation (reviewed by Schwob et al., 1980; Nadler, 1981; Riba-Bosch and Pérez-Clausell, 2004). We injected Sprague-Dawley rats of 150–170 g, b.w., with 12 mg/kg of kainate and examined neuronal loss exclusively in the olfactory cortex and hippocampus from 6 h to 7 days after treatment. These regions were selected because they show a different vulnerability to kainate-induced damage. We arbitrarily subdivided kainate-injected animals into high responders and low responders on the basis of the seizure severity score and EEG activity. Neurons of the olfactory cortex were highly vulnerable to kainate-induced seizures, and appeared to be severely damaged both in high and low responder animals. Hippocampal neurons were less vulnerable, and damage developed only in high responder animals. Under our experimental conditions, hippocampal damage was prominent in the CA1 region, at least within the time frame selected for our analysis (i.e., 7 days). Using Sprague-Dawley rats of 300–350 g, b.w., Narkilahti and Pitkänen (2005) found a greater damage in the CA1 region and the hilus than in the CA3 region at 48 h, but no difference among the three subregions at 7 days following systemic kainate injection (12 mg/kg, i.p.). Others have found neuronal loss and granule cell axon reorganization in the hippocampus dentate gyrus following different paradigms of kainate injection (Buckmaster and Dudek, 1997; Buckmaster and Jongen-Relo, 1999). An investigation of the different pattern of neuronal loss between our study and other studies with systemic injection of kainate is beyond the scope of this article.
We found a remarkable correlation between induction of Dkk-1 and neuronal death in rats developing seizures in response to kainate injection. Dkk-1 was consistently expressed in the olfactory cortex of high responder rats in which neuronal death is >90%, and was not detected under conditions in which seizures did not cause neuronal damage, e.g. in the hippocampus of low responder rats. Dkk-1 expression always anticipated neuronal loss with a lag time that was strictly related to the development of neuronal damage. For example, when compared to high responder rats, Dkk-1 expression and neuronal loss in the olfactory cortex were synchronously delayed in low responder rats. This suggests that the harmful consequences of synaptic hyperactivity cumulate with time until they reach the threshold for the induction of Dkk-1 in neurons destined to die. Different mechanisms have been implicated in the pathophysiology of seizure-induced neuronal damage, including an increased Ca2+ influx (Schanne et al., 1979), through kainate and NMDA channels (Lafon-Cazal et al., 1993), formation of radical oxygen species (Bruce and Baudry, 1995; Ueda et al., 1997), and mitochondrial dysfunctions (Chuang et al., 2004). These mechanisms converge in causing DNA damage, and the ensuing expression of p53 may activate transcription of the Dkk-1 gene (Wang et al., 2000). Accordingly, the lack of p53 prevents Dkk-1 induction in cultured neurons challenged with β-amyloid (Caricasole et al., 2004). Intracerebroventricular injection of Dkk-1 antisenses induced a near-to-complete protection against seizure-induced neuronal death in the olfactory cortex and hippocampus. A similar extent of neuroprotection was observed in vivo models of ischemic neuronal death and in cultured neurons challenged with excitotoxins (Cappuccio et al., 2005). We suggest that induction of Dkk-1 contributes to the execution of a death program induced by a variety of toxic insults, including a sustained synaptic hyperactivity. The reduction of nuclear β-catenin levels found in the hippocampus of kainate-injected animals suggests that Dkk-1 inhibits the canonical Wnt pathway, limiting the amount of β-catenin available for nuclear translocation (Levina et al., 2004). This would deprive neurons from the activation of a gene program that is essential for their survival (Willert and Nusse, 1998). A role for Wnt inhibition in the development of seizure-induced neuronal damage was confirmed by experiments in which kainate was injected to animals pretreated with lithium. Lithium is an established inhibitor of GSK-3β (Klein and Melton, 1996) and, therefore, can rescue the Wnt pathway acting downstream of the Dkk-1 blockade. Remarkably, lithium was neuroprotective without affecting the extent, duration, and EGG characteristics of kainate-induced seizures.
Whether and to what extent an early inhibition of the Wnt pathway contributes to the pathophysiology of hippocampal sclerosis in humans remains to be established. We have consistently found a strong expression of Dkk-1 in the hippocampus of patients with temporal lobe epilepsy associated with hippocampal sclerosis. Dkk-1 was absent or expressed at very low levels in nonepileptic tissue or in samples from patients with partial complex seizures without evidence of hippocampal sclerosis, except in the hippocampus from one epileptic patient in which perhaps neurons were destined to die. This strongly confirms that Dkk-1 is induced in relation to neuronal death and not simply as a result of synaptic hyperactivity.
Taken together, our data raise the attractive possibility that Dkk-1 antagonists (currently under development) or GSK-3β inhibitors limit the extent of neuronal damage associated with temporal lobe epilepsy independently of their effect on seizure activity. One of these inhibitors is valproic acid (Loscher, 2002). It will be interesting to examine whether valproic acid is neuroprotective independently of its established antiepileptic activity.