Address correspondence and reprint requests to Rodrigo A. Cunha, Center for Neuroscience of Coimbra, Institute of Biochemistry, Faculty of Medicine, University of Coimbra, 3004-504 Coimbra, Portugal. E-mail: firstname.lastname@example.org.
Seizures early in life cause long-term behavioral modifications, namely long-term memory deficits in experimental animals. Since caffeine and adenosine A2A receptor (A2AR) antagonists prevent memory deficits in adult animals, we now investigated if they also prevented the long-term memory deficits caused by a convulsive period early in life. Administration of kainate (KA, 2 mg/kg) to 7-days-old (P7) rats caused a single period of self-extinguishable convulsions which lead to a poorer memory performance in the Y-maze only when rats were older than 90 days, without modification of locomotion or anxiety-like behavior in the elevated-plus maze. In accordance with the relationship between synaptotoxicity and memory dysfunction, the hippocampus of these adult rats treated with kainate at P7 displayed a lower density of synaptic proteins such as SNAP-25 and syntaxin (but not synaptophysin), as well as vesicular glutamate transporters type 1 (but not vesicular GABA transporters), with no changes in PSD-95, NMDA receptor subunits (NR1, NR2A, NR2B) or α-amino-3-hydroxy-5-methylisoxazole-4-propionate receptor subunits (GluR1, GluR2) compared with controls. Caffeine (1 g/L) or the A2AR antagonist, KW6002 (3 mg/kg) applied in the drinking water from P21 onwards, prevented these memory deficits in P90 rats treated with KA at P7, as well as the accompanying synaptotoxicity. These results show that a single convulsive episode in early life causes a delayed memory deficit in adulthood accompanied by a glutamatergic synaptotoxicity that was prevented by caffeine or adenosine A2AR antagonists.
During development, the brain undergoes marked plastic changes which format its wiring and determine its potential for future experience-dependent plastic changes, i.e. the developing brain expresses a marked metaplasticity (Abraham and Bear 1996). An abnormal modification of firing of the developing brain, as occurring during convulsions, causes minor immediate modifications but has potential detrimental consequences later in life (Stafstrom 2002; Holmes 2005). In fact, in animal models, a single convulsive period early in life causes delayed memory deficits in adulthood (Stafstrom 2002; Holmes 2005). For instance, the administration of kainate, which models temporal lobe epilepsy in adult rodents, causes minor immediate morphological or functional modifications in pups (Stafstrom 2002; Holmes 2005), but triggers an impairments of spatial memory when animals are 80–100 days old (Lynch et al. 2000; Sayin et al. 2004). The mechanisms underlying this metaplasticity are not resolved (Holmes 2005). But more important than understanding the mechanisms of convulsions-induced delayed memory deficits is the possibility of devising novel strategies to preserve memory, which may also offer some additional insights into key mechanistic processes.
One emerging candidate to manage memory impairment under different noxious conditions is caffeine. Caffeine is the most widely consumed psychoactive substance and alleviates cognitive impairment in both humans and animals, namely upon aging or upon Alzheimer’s disease (Cunha 2008; Takahashi et al. 2008). Furthermore, caffeine also affords protection upon CNS injury (Cunha 2005; Chen et al. 2007). The only known molecular targets of caffeine at non-toxic doses are adenosine receptors, mainly A1 and A2A receptors (A2AR) (Fredholm et al. 2005). A2AR blockade seems to be primarily involved in these neuroprotective effects of caffeine since the prevention by caffeine of memory deficits is mimicked by antagonists of A2AR but not A1R (Prediger et al. 2005; Dall’Igna et al. 2007).
In the present study, we investigated if chronic treatment with caffeine or with a selective A2AR antagonist could prevent the memory deficits found in adult animals that were exposed to an episode of convulsions early in life. Since memory deficits in different neuropsychiatric conditions are accompanied by dysfunction and loss of synapses (Coleman et al. 2004; Silva et al. 2007), we also investigated if a convulsive episode early in life caused a modification of synaptic markers in adulthood while testing if there was a particular modification of pre- and post-synaptic markers of glutamatergic synapses.
Materials and methods
Wistar rats (Charles-River, Barcelona, Spain) were maintained under controlled environment (23 ± 2°C, 12 h-light/dark cycle, free access to food and water) and used according to EU guidelines (86/609/EEC) with care to minimize the number of animals and their suffering.
Kainate-induced convulsion and drug administration
Male young pups (7 days old, P7), bred at the Center’s animal house by trained caretakers, were separated from their dams. A single convulsive episode was induced by intra-peritoneal administration of kainic acid (KA) 2 mg/kg (Lynch et al. 2000; Sayin et al. 2004) prepared in saline solution. Control littermates were injected with saline solution. After spontaneous termination of convulsions, within 3 h after their onset, KA-injected and control rats were returned to their dams.
Caffeine was added to the drinking water using a dose (1 g/L) previously found to afford neuroprotection (Duarte et al. 2009). Caffeine was introduced at P21 (weaning day) and continuously supplied until rats were killed. Non-treated rats (KA- or saline-treated) drank water only. Weight and water/caffeine consumption were continuously monitored, which allowed us to estimate the daily amount of caffeine consumption as 69 ± 3 mg/kg/day (consumed in a continuous intermittent mode) without correction for spillage (estimated to be ca. 5%). Measurement of serum caffeine concentration was done by reverse-phase HPLC using 40% (v/v) methanol/water as eluent (flow rate of 0.8 mL/min) with UV detection at 257 nm. The average caffeine intake was similar in both KA- and saline-treated rats through the treatment period, which displayed similar serum caffeine concentrations (21.4 ± 2.8 μM, n = 6) at the time of killing. This corresponds to the plasma concentration reached by consumption of 5–6 cups of coffee in humans (e.g. Lelo et al. 1986).
Chronic intake of the selective A2AR antagonist, KW6002 (istradefylline, see Kase et al. 2003), also begun at P21. Everyday, a 1 mg/L KW6002 suspension (prepared as previously described, see Hockemeyer et al. 2004) in a vehicle solution (0.9% saline and 0.4% methylcellulose) was prepared and the adequate volume of the suspension was added to the drinking water to obtain the desired dose of KW6002 (3 mg/kg). This solution was left for a period of 2.5 h to ensure that rats consumed an amount of 3 mg/kg/day of KW6002 (again with a spillage error of ca. 5%). Non-treated rats (KA- or saline-treated) drank water supplemented with vehicle solution.
Behavioral experiments were conducted between 9:00 am and 4:00 pm (light phase). Locomotor and exploratory behavior were monitored in an open-field apparatus, as previously described (Dall’Igna et al. 2007), when rats were 28, 58, and 88 days old. On the subsequent days (29, 59, and 89 days), rats were submitted to the elevated-plus maze task to evaluate their anxiety status (Kaster et al. 2004). Finally, on days 30, 60, and 90, rats performed a memory task in a Y-maze apparatus with an inter-trial interval (ITI) of 2 min and with a 2 h ITI 1 week later. We used an adapted version of the Y-maze designed to measure spatial recognition memory (Dellu et al. 1997). The three arms of the Y-maze were randomly designated: start arm, in which rats started to explore (always open), novel arm, which was blocked during the first trial, but open during the second trial, and other arm (always open). The Y-maze task consisted of two trials separated by an ITI to assess response to novelty (2 min ITI) and spatial recognition memory (2 h ITI) (Dellu et al. 1997). During the first trial (training, 5 min), rats were allowed to explore only two arms (start and other arm), with the third arm (novel arm) closed. For the second trial (after ITI), the rat was placed back in the same starting arm, with free access to all three arms for 5 min. The number of entries in each arm was determined and data expressed as percentage of total entries during the 5 min (time spent in each arm was also counted yielding similar results but is not displayed for clarity of Figures).
Western blot analysis
After all behavior tests, some rats were deeply anesthetized under halothane atmosphere before being killed by decapitation for preparing Percoll-purified hippocampal synaptosomes as described (Rebola et al. 2005; Canas et al. 2009). Western blot analysis was carried out in these synaptosomal membranes as previously described (Rebola et al. 2005; Canas et al. 2009), except for glial fibrillary acidic protein (GFAP) which was evaluated in total hippocampal membranes (Rebola et al. 2005). The antibodies were against SNAP-25 (1 : 5000 dilution; from Sigma, Sintra, Portugal), syntaxin (1 : 5000, Sigma), synaptophysin (1 : 2000, Sigma), vesicular glutamate transporters type 1 (vGluT1, 1 : 5000, Chemicon, PG-Hitec, Lisbon, Portugal), vesicular GABA transporter (vGAT, 1 : 1000, Calbiochem, VWR, Lisbon, Portugal), post-synaptic density 95 kDa protein (PSD-95, 1 : 20 000, Upstate Biotechology, PG-Hitec, Lisbon, Portugal), NMDA receptor subunit 1 (NR1, 1 : 400, Chemicon), NMDA receptor subunit 2A (NR2A, 1 : 800, Chemicon), NMDA receptor subunit 2B (NR2B, 1 : 200, BD Biosciences, PG-Hitec, Lisbon, Portugal), α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) receptor subunit 1 (GluR1, 1 : 400, Upstate Biotechnology), AMPA receptor subunit 2 (GluR2, 1 : 400, Chemicon) or GFAP (1 : 1000, Cell Signalling, Izasa, Lisbon, Portugal). Re-probing quantifying α-tubulin (1 : 10 000, Sigma) immunoreactivity confirmed that similar amounts of protein were applied to the gels (Rebola et al. 2005; Canas et al. 2009). We also confirmed, by loading different amounts of protein in the same gel, that we were working under non-saturating conditions enabling to effectively probe changes in the density of each tested protein (Rebola et al. 2005; Canas et al. 2009).
Membrane binding assays
Binding assays were carried out as previously described (Cunha et al. 2006) in membranes from hippocampal synaptosomes using supra-maximal concentrations (6 nM) of selective antagonists of either A1R (3H-1,3-dipropyl-8-cyclopentylxanthine, 3H DPCPX; specific activity of 109.0 Ci/mmol; from DuPont NEN, ILC, Lisbon, Portugal) or A2AR (3H-7-(2-phenylethyl)-5-amino-2-(2-furyl)-pyrazolo-[4,3-e]-1,2,4-triazolol[1,5c]pyrimidine, 3H-SCH58261; specific activity of 77 Ci/mmol; prepared by Amersham and offered by Dr E. Ongini, Shering-Plough, Italy).
Immunocytochemistry in purified nerve terminal
Immunocytochemical analysis of purified hippocampal nerve terminals was performed as previously described (Rodrigues et al. 2008). The platted nerve terminals were labeled with goat anti-adenosine A2AR (1 : 500; Santa Cruz Biotechnology, Heidelberg, Germany) or with rabbit anti-adenosine A1R antibody (1 : 400; from Affinity Bioreagents, PG-Hitec, Lisbon, Portugal), together with mouse anti-synaptophysin antibody (1 : 200; Sigma) and visualized in a Zeiss Axiovert 200 inverted fluorescence microscope equipped with a CCD camera and analyzed with ImageJ 1.37v (NIH, Bethesda, MD, USA). Each coverslip (three per animal) was analyzed by counting five different fields and in each field a total amount of at least 100 individualized nerve terminals.
The preparation of brain sections was carried out as previously described (Cunha et al. 2006) and always analyzed by three independent researchers. The general neuronal morphology was evaluated with Cresyl Violet staining of Nissl bodies, as previously described (Lopes et al. 2003). Hippocampal astrocytes were evaluated by immunohistochemical detection of GFAP, a marker of activated astrocytes (Pekny and Nilsson 2005). Blocked brain sections were incubated for 72 h at 4°C with Cy3-conjugated anti-GFAP mouse monoclonal antibody (from Sigma; 1 : 1000 dilution containing 0.2% Triton X-100 and 10% goat serum). After washing, sections were dehydrated and cleared in xylol and mounted on slides using Vectaschield H-1400 mounting medium (Vector Laboratories, Batista Marques, Lisbon, Portugal). The numbers of immuno-staining-positive cells in each hippocampal region was averaged from four hippocampal sections per rat.
Data are mean ± SEM of n animals. Significance (p < 0.05) was assessed by one-way Student’s t-test or by a one-way ANOVA followed by Bonferroni’s test to compare data of the four experimental groups.
A single episode of convulsions in early life causes memory impairment in adult rats
Kainate (KA, 2 mg/kg, i.p.) caused a convulsive-like activity in 7-day-old pups (P7); this consisted of hyperactive ‘bicycling’ movements of all extremities with opisthonic arching of the back and tonic limbic extension, occurring with a fast onset (ca. 10 min), lasting 80–210 min, and disappearing spontaneously, as previously reported (Lynch et al. 2000; Sayin et al. 2004).
We then tested if this single convulsive episode at P7 modified behavior later in adolescence and adulthood. At P30 or P60, KA-injected rats (at P7) displayed a behavior profile indistinguishable from that of saline-injected rats (Fig. 1, third and second rows). However, at P90, KA-injected rats displayed significantly (p < 0.05) poorer performance in a spatial memory version of the Y-maze test compared with saline-injected rats (Fig. 1, top left panel, whereas locomotion and anxiogenic-like behavior were not significantly different (p > 0.05) from saline-injected rats (Fig. 1, first row).
Caffeine and A2AR blockade prevent convulsions-induced memory impairment
To test if consumption of caffeine or selective A2AR antagonist prevented this delayed memory deficit (only observable at P90, but not at P30 or P60) caused by a convulsive episode early in life, rats injected with either KA- or saline at P7 were divided into two groups: one group was exposed to caffeine (1 g/L) or to the selective A2AR antagonist, KW6002 (3 mg/kg, see Kase et al. 2003), whereas the non-treated group only drank water from P21 (weaning) onwards.
Caffeine consumption prevented memory impairment of KA-treated rats at P90, whereas caffeine was devoid of effect in saline-injected rats (Fig. 2, upper left panel). Furthermore, locomotion and anxiety behaviors were similar in all four groups (Fig. 2, upper panels).
Likewise, memory impairment displayed at P90 by KA-treated rats was abrogated by KW6002 consumption from P21 onwards (Fig. 2, bottom left panel), whereas KW6002 did not modify behavior of saline-injected rats. Also, locomotion and anxiety behaviors were similar (p > 0.05) in all four groups (Fig. 2b, bottom panels).
Impact of convulsions early in life on the set-up of adenosine receptors in adulthood
We next investigated the status of hippocampal adenosine receptors of KA-treated rats at P90. Binding of the A2AR antagonist 3H-SCH58261 was increased by 65.3 ± 4.2% (p < 0.05; Fig. 3b), whereas binding of the A1R antagonist 3H- DPCPX was decreased by 24.0 ± 2.4% in P90 rats injected with KA at P7 compared with saline-injected rats (p < 0.05; Fig. 3a). Furthermore, the percentage of synaptophysin-immunopositive terminals endowed with A1R immunoreactivity was similar (Fig. 3d), whereas the number of synaptophysin-immunopositive terminals endowed with A2AR immunoreactivity increased in P90 rats injected with KA at P7 compared with saline-injected rats (Fig. 3c and e). This indicates that KA-induced convulsions at P7 led to an increase of nerve terminals endowed with A2AR, whereas the number of nerve terminals endowed with A1R was preserved; however, the density of A1R per terminal might have decreased, since receptor binding analysis revealed a global decrease of A1R in hippocampal synaptosomal membranes (Fig. 3a).
Convulsions early in life lead to features of synaptotoxicity in adulthood
We next investigated if early-life convulsions-induced memory impairment in adult rats was associated with synaptic degeneration, which has been implicated in Alzheimer’s disease-associated memory impairment (Selkoe 2002; Coleman et al. 2004). We found a reduced immunoreactivity of SNAP-25 (−21.4 ± 3.7%, p < 0.05) and syntaxin (−20.2 ± 3.0%, p < 0.05) but not of synaptophysin (−6.3 ± 3.0%, p > 0.05) in hippocampal membranes of P90 rats treated with KA at P7 (Fig. 4a–c), suggesting the occurrence of synaptic degeneration. There was also a reduced immunoreactivity of vGluT1 (-27.2 ± 3.3%, p < 0.05) but not of vGAT (2.3 ± 2.6%, p > 0.05), suggesting the occurrence of glutamatergic rather than GABAergic synaptic degeneration (Fig. 4d and e). In contrast, the immunoreactivity of PSD-95 (Fig. 4f), NR1 (Fig. 4g), NR2A (Fig. 4h) or NR2B subunits of NMDA receptors (Fig. 3i) and of GluR1 (Fig. 4j) or GluR2 subunits of AMPA receptors (Fig. 4k) was not modified (p > 0.05).
Semi-quantitative immunohistochemistry of the astrocytic protein marker GFAP (Pekny and Nilsson 2005) revealed a similar (p > 0.05) density of GFAP-positive cells in the hippocampus of P90 rats injected with KA at P7, compared with saline-treated rats in the three hippocampal regions (Figs 5b and c), which was confirmed by western blot analysis (Fig. 5d), suggesting a lack of evident glial activation.
Caffeine and A2AR antagonist prevent adulthood synaptotoxicity upon convulsions early in life
We next investigated if caffeine or KW6002 prevented this synaptotoxicity at P90 upon KA administration at P7. As illustrated in Fig. 6, the consumption of caffeine or KW6002 from P21 onwards prevented the decreased density of SNAP-25 (p < 0.05) and of syntaxin (p < 0.05) in hippocampal membranes of P90 rats that were subject to kainate-induced convulsion at P7; likewise, the decrease of vGluT1 was also prevented (p < 0.05) by both caffeine and KW6002 (Fig. 6), which were devoid of effects on vGAT density that was similar (p > 0.05) in all four groups (Fig. 6).
The present results indicate that a single period of KA-induced convulsions in early life (P7) triggers selective memory deficits later, only in adulthood (P90), which are accompanied by a loss of pre-synaptic markers, in particular of glutamatergic terminals, without modification of post-synaptic markers, astrogliosis or global organization of principal neurons (Cresyl violet staining) in the hippocampus. Furthermore, chronic consumption of caffeine or of a selective A2AR antagonist prevents memory deficits and concurrent synaptotoxicity present in adult rats that were subject to a single period of convulsions early in life.
The first conclusion of this study was that a period of convulsions at P7 led to selective memory deficits, which is delayed in time, appearing only in adulthood. We now show that the spatial memory deficits were observed only at P90 as observed by others (Lynch et al. 2000; Sayin et al. 2004), but are not evident in late adolescence (P30) or early adulthood (P60), an aspect which was not systematically addressed in previous studies. The extrapolated relevance of these findings for humans is provocative: although it is still debatable if early-life convulsions are associated with poorer memory performance, this was only evaluated in late adolescence (Shinnar and Hauser 2002; Vingerhoets 2006). This leaves room to question whether deficits might only become evident later in adulthood.
The second conclusion of this study is related to the possible neurochemical traits underlying the observed memory deficits. Adult animals subject to KA-induced convulsions in early life display minor morphological modifications compared with control animals (Lynch et al. 2000; Stafstrom 2002; Holmes 2005; Cornejo et al. 2007), whereas they display neurophysiological modifications, such as modified synaptic plasticity (Lynch et al. 2000; Cornejo et al. 2007). Interestingly each group proposed different neurochemical explanations to interpret these neurophysiological changes, namely enhanced efficiency of inhibitory transmission (Lynch et al. 2000), enhanced efficiency of glutamatergic synapses (Cornejo et al. 2007) or persistent astrogliosis (Somera-Molina et al. 2007). In the present study, we failed to find evidence for persistent hippocampal astrogliosis since neither GFAP density, nor the number of GFAP-labeled astrocytes, nor the morphology of GFAP-labeled elements were modified in the hippocampus of rats exposed to early-life convulsions. However, we provide the first direct demonstration of pre-synaptic modifications accompanying memory deficits in adult rats that were subject to a convulsive period in early life. Thus, we observed a decrease of two synaptic markers, SNAP-25 and syntaxin. This synaptotoxicity has also been proposed to be a primary and crucial feature responsible for memory impairment occurring in mild cognitive impairment (Scheff et al. 2007) and Alzheimer’s disease (Selkoe 2002; Coleman et al. 2004), since synaptotoxicity is the only morphological parameter that correlates with memory impairment (Selkoe 2002; Coleman et al. 2004). Furthermore, features of synaptotoxicity are characteristic of several other brain conditions where memory impairment is also present, such as aging (Canas et al. 2009), Huntington’s (Li et al. 2001) or prions’ diseases (Ferrer 2002), HIV infection (Garden et al. 2002) or schizophrenia (Glantz et al. 2006). Most interestingly, we found a reduction of a marker of glutamatergic synapses (vGluT1), whereas a marker of GABAergic synapses (vGAT) was preserved, indicating selective changes of glutamatergic rather than GABAergic synapses. This is in remarkable agreement with recent results indicating that memory-related synaptotoxicity might occur particularly in glutamatergic terminals, since the density of vesicular glutamate transporters was found to be decreased in cortical regions of memory-impairment individuals with Alzheimer’s disease (Kirvell et al. 2007), as well as in animal models of Alzheimer’s disease (Bell et al. 2006; Minkeviciene et al. 2008). Thus, as occurs for Alzheimer’s disease, it is likely that the reduced density of hippocampal synaptic proteins, particularly of glutamatergic terminals, may contribute to the memory impairment found in adult rats that were exposed to convulsions early in life. Interestingly, we mostly found modifications of pre-synaptic markers of glutamatergic synapses rather than changes in the density of different subunits of ionotropic glutamate receptor, in spite of the fact that the convulsive period was applied precisely at the time where the maturation of receptors in glutamatergic synapses is taking place (reviewed in Jensen 2002).
The likely involvement of synaptotoxicity in the mechanism of memory impairment in adulthood caused by early-life convulsions is further supported by the main finding of the present work, i.e. that long-term caffeine consumption prevented both memory impairments and loss of nerve terminal markers in the hippocampus of adult rats that were exposed to convulsions early in life. This is paralleled by the ability of caffeine to prevent memory deficits found in aging, in Alzheimer’s and Parkinson’s diseases and in attention deficit and hyperactivity disorders (Cunha 2008; Takahashi et al. 2008), suggesting that caffeine can interfere with key mechanisms of memory dysfunction (Cunha 2008). The only known molecular targets for caffeine at non-toxic concentrations are adenosine receptors, which are antagonized by caffeine (Fredholm et al. 2005). The prevention of memory deficits by caffeine likely results from A2AR antagonism since the beneficial effects of caffeine on memory deficits upon aging or Alzheimer’s disease are mimicked by selective A2AR antagonists (Prediger et al. 2005; Dall’Igna et al. 2007). Accordingly, we now observed that the selective A2AR antagonist, KW6002, also prevented both memory impairment and loss of nerve terminal markers in the hippocampus of adult rats that were exposed to a convulsive period in their early life. Interestingly, A2AR antagonism not only abrogates memory dysfunction but also affords robust neuroprotection against brain damage (Cunha 2005; Chen et al. 2007). This further supports the involvement of synaptotoxicity in memory dysfunction since A2AR have a predominant synaptic localization in the hippocampus (Rebola et al. 2005), where they control synaptotoxicity (Cunha et al. 2006; Silva et al. 2007), which is the most evident morphological change found in the hippocampus of adult rats that were exposed to a convulsive period early in their life. These synaptic A2AR undergo a gain of function in noxious brain conditions (Cunha 2005; Fredholm et al. 2005). Accordingly, we now found an increase in the density of synaptic A2AR in the hippocampus of adult rats that were exposed to a convulsive period early in their life. However, the mechanism by which A2AR control synaptotoxicity in glutamatergic synapses is still unclear, albeit the control of NMDA receptors (Rebola et al. 2008), calcium loading (Gonçalves et al. 1997) and synaptic mitochondria (Silva et al. 2007) are likely candidates.
In conclusion, the present study shows that a convulsive period early in life causes selective impairment of memory performance later in adulthood, which is accompanied by a loss of synaptic markers, in particular of glutamatergic terminals. Furthermore, chronic consumption of caffeine or of A2AR antagonists starting in adolescence prevented this delayed memory deficit and accompanying synaptotoxicity. These observations widen the prophylactic interest in caffeine and A2AR antagonists to manage conditions associated with memory deterioration.
The authors are grateful to Paula M. Canas, Ana Patrícia Simões, Carla G. Silva, Gabriele Ghisleni and Manuella P. Kaster for their help in carrying out some of the experiments. GPC acknowledges the support and help of Lisiane O. Porciúncula and Carla Bonan. The dedicated and competent help of Alexandre Pires to handle the animals is especially acknowledged. This work was supported by FCT (PTDC/SAU-NEU/74318/2006) and was made possible by a joint Portuguese-Brazilian grant (CAPES-GRICES).