INTRODUCTION

Lithium (Li+) has been used to treat bipolar disorder (BD) for about 50 years, but its mechanism of action is not agreed on (Barchas et al, 1994; Cade, 1999). One suggestion has been that it acts in part by inhibiting glutamatergic neurotransmission, particularly via N-methyl-D-aspartate (NMDA) receptors (NMDARs) (Bauer et al, 2003; Farber et al, 2002; Krystal et al, 2002; Stewart and Reid, 2002; Zarate et al, 2003). Supporting this is evidence of disturbed markers of NMDA functioning in the BD brain (Itokawa et al, 2003; Mundo et al, 2003; Scarr et al, 2003), and data that drugs effective against BD, such as lamotrigine, carbamazepine and valproic acid, have direct or indirect NMDA antagonist properties (Farber et al, 2002; Hough et al, 1996; Kubota et al, 1994; McIntyre et al, 2004; Zeise et al, 1991).

Ionotropic NMDARs, cation channels with a high permeability to Ca2+ when activated, consist of heterotetrameric assemblies of NR1, NR2A-D, and NR3 subunits. The subunits are assembled cotranslationally in the endoplasmic reticulum, from where they are transported and then inserted at glutamatergic synapses. Release of NMDARs from the endoplasmic reticulum requires phosphorylation of NR1 and NR2 by protein kinases A and C (PKA and PKC), or by serine or tyrosine kinases (Scott et al, 2003; Wenthold et al, 2003). Binding of glutamate or NMDA to synaptic NMDARs allows Ca2+ into the cell to activate Ca2+-dependent enzymes such as calcium/calmodulin-dependent protein kinase II (CaMKII), PKC, phospholipase Cγ, and phospholipase A2 (PLA2) (Colbran, 2004; Fukunaga et al, 1992; Gurd and Bissoon, 1997; Weichel et al, 1999; Wenthold et al, 2003).

Exposure of neurons from different brain regions to NMDA is reported to activate PLA2 through the Ca2+ mechanism, to release the second messenger, arachidonic acid (AA, 20 : 4 n-6), from membrane phospholipid (Dumuis et al, 1988; Kim et al, 1995; Kolko et al, 1999, 2003; Lazarewicz et al, 1990; Pellerin and Wolfe, 1991; Sanfeliu et al, 1990; Tapia-Arancibia et al, 1992; Tencé et al, 1994; Weichel et al, 1999). The AA release could be blocked by the nonspecific PLA2 inhibitor, mepracrine, or in the rat hippocampus by bilobalide, a constituent of Ginkgo biloba, and by a specific inhibitor of cPLA2 (cytosolic PLA2) (see below) (Sanfeliu et al, 1990; Tapia-Arancibia et al, 1992; Tencé et al, 1994; Weichel et al, 1999). In contrast, NMDA did not stimulate AA release in hippocampal astroglia (Sanfeliu et al, 1990).

Optical methods have been used to image the Ca2+ signal in response to exogenous NMDA in brain slices or cell culture (Hashimoto et al, 2002; Ismailov et al, 2004; Mainen et al, 1999; Takita et al, 1997). However, no method exists, as far as we know, to image the Ca2+ signal or its downstream effects in the brain in vivo. In this paper, we present such a method, which extends our published method of imaging brain PLA2 activation in unanesthetized rats given drugs that act at receptors coupled a G-protein to PLA2 (Basselin et al, 2005a, 2005b, 2003; Bhattacharjee et al, 2005; DeGeorge et al, 1991; Qu et al, 2003)—cholinergic muscarinic M1,3,5, dopaminergic D2-like and serotonergic 5-HT2A/2C receptors (Axelrod, 1995; Bayon et al, 1997; Felder et al, 1990; Vial and Piomelli, 1995).

There are three classes of brain PLA2 enzymes: AA-selective Ca2+-dependent cPLA2, sPLA2 (secretory PLA2), and iPLA2 (Ca2+-independent PLA2) (thought to be selective for docosahexaenoic acid (DHA, 22 : 6 n-3) rather than AA). All have been identified in neurons (Kishimoto et al, 1999; Strokin et al, 2004; Yang et al, 1999; Yegin et al, 2002). cPLA2 has been localized at postsynaptic neuronal membranes in the brain (Basavarajappa et al, 1998; Ong et al, 1999; Pardue et al, 2003) and is activated by 300 nM to 1 μM Ca2+ (Clark et al, 1995), in the physiological range of intracellular Ca2+ during neuronal activation (Ismailov et al, 2004). sPLA2 is activated at much higher Ca2+ concentrations, 16–18 mM (Dennis, 1994), and is localized in presynaptic vesicles that are released by exocytosis during membrane depolarization (Matsuzawa et al, 1996; Wei et al, 2003).

In elaborating our in vivo fatty acid imaging method when stimulating receptors coupled to PLA2 (see above), we concluded that a fraction of the AA that is released following PLA2 activation is rapidly reincorporated into phospholipid, whereas the remainder is lost by conversion to eicosanoids or other products, or by β-oxidation (Rapoport, 2001, 2003; Robinson et al, 1992). Unesterified AA in plasma rapidly replaces the quantity lost, as AA is nutritionally essential and cannot be synthesized de novo in vertebrate tissue or elongated significantly from its precursor, linoleic acid (18 : 2 n-6) (DeMar et al, 2004b; Holman, 1986; Washizaki et al, 1994). Replacement is proportional to PLA2 activation and can be imaged by injecting radiolabeled AA intravenously, then measuring regional brain radioactivity by quantitative autoradiography. A regional AA incorporation coefficient k* (regional brain radioactivity/integrated plasma radioactivity), calculated in this way, has been shown to be independent of changes in cerebral blood flow and to represent the plasma-derived AA reincorporated into brain phospholipid (Basselin et al, 2003; Chang et al, 1997; DeGeorge et al, 1991; DeMar et al, 2004a; Rapoport, 2001, 2003; Robinson et al, 1992).

We thought it important to see if we could image brain AA signaling in response to NMDA, and if chronic LiCl would attenuate this signaling. In this study, we injected radiolabeled AA intravenously into unanesthetized rats to determine k* for AA in each of 83 brain regions in response to i.p. saline, and to each of two i.p. doses of NMDA, 25 and 50 mg/kg, neither of which is reported to produce convulsions although capable of inducing paroxysmal EEG activity (Ormandy et al, 1991). We also imaged regional k* responses to the selective noncompetitive NMDA antagonist, MK-801 ((5R,10S)-(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine hydrogen maleate) (Wong et al, 1986, 1988), given at a dose of 0.3 mg/kg i.p. alone or prior to NMDA administration. Alone, this dose is reported to produce ataxia, hyperlocomotion, and stereotypies (Dai et al, 1995) accompanied by cortical spike-wave discharges (Feinberg et al, 1995). Finally, we measured k* for AA in response to NMDA in animals fed a control or LiCl diet for 6 weeks. The LiCl diet produces plasma and brain lithium concentrations of about 0.8 mM, reported to be therapeutic in BD (Basselin et al, 2003; Bosetti et al, 2002b; Calabrese et al, 1995; Chang et al, 1996). An abstract of part of this work has been presented (Basselin et al, 2004).

MATERIALS AND METHODS

Animals and Diets

Experiments were conducted following the ‘Guide for the Care and Use of Laboratory Animals' (National Institute of Health Publication No. 86-23) and were approved by the Animal Care and Use Committee of the National Institute of Child Health and Development. Male Fischer CDF (F-344)/CrlBR rats (2 months old) (Charles River Laboratories, Wilmington, MA) were fed a control or LiCl diet as previously described (Basselin et al, 2003).

Drugs

NMDA (25 or 50 mg/kg, RBI Signaling Innovation, Sigma-Aldrich, Natick, MA), (+)-MK-801 hydrogen maleate (0.3 mg/kg, RBI Signaling Innovation) or 0.9% NaCl (saline) was administered i.p. [1-14C]AA in ethanol (53 mCi/mmol, >98% pure, Moravek Biochemicals, Brea, CA) was evaporated and resuspended in HEPES buffer, pH 7.4, containing 50 mg/ml BSA, as described (DeGeorge et al, 1989). To confirm tracer purity, gas chromatography was performed after the [1-14C]AA was converted to a methyl ester using 1 ml of 1% H2SO4 in anhydrous methanol (Makrides et al, 1994).

Surgical Procedures and Tracer Infusion

Surgery was performed as described (Basselin et al, 2003; Chang et al, 1996). A rat was wrapped loosely in a fast-setting plaster cast taped to a wooden block, and allowed to recover from anesthesia for 3–4 h. Body temperature was maintained at 36.5°C using a feedback-heating device. Arterial blood pressure and heart rate were measured 8 and 28 min after NMDA and MK-801 injection, respectively. Behavior was monitored immediately after drug administration until the end of the experiment. Repeated cycles consisting of an activity followed by a calm period were produced by NMDA. We measured the mean duration of each of these periods, as well as of the net cycling period.

At 10 min after NMDA or 30 min after MK-801 injection [1-14C]AA (170 μCi/kg, 2 ml, 400 μl/min) was infused for 5 min via the femoral catheter using an infusion pump (Harvard Apparatus Model 22, Natick, MA). For studies with both drugs, MK-801 was administered 30 min prior to NMDA, which in turn was given 10 min before [1-14C]AA. Timed arterial blood samples were collected during radiotracer infusion, to time of death at 20 min, when the rat was killed using an overdose of sodium pentobarbital (100 mg/kg i.v.) and immediately decapitated. The brain was removed and quickly frozen in 2-methylbutane at −40°C, stored at −80°C, and later sectioned for autoradiography.

Chemical Analysis

Plasma was separated from arterial blood by centrifugation, and lipids were extracted using a modification of Folch's method (DeGeorge et al, 1991; Folch et al, 1957). Total radioactivity in the organic phase was measured by liquid scintillation counting.

Quantitative Autoradiography

Quantitative autoradiography was performed as described previously (Basselin et al, 2003). A total of 83 brain regions were identified by comparing autoradiographs of coronal brain sections with a coronal atlas of the rat brain (Paxinos and Watson, 1987). The average of three bilateral optical density readings for each brain region from each animal was used to calculate regional radioactivity. Regional brain incorporation coefficients of AA from plasma into the stable brain lipid compartment, k* (ml/s/g brain), were calculated as

where cbrain* (20 min) nCi/g equals brain radioactivity at the time of death (20 min), cplasma* nCi/ml equals plasma AA radioactivity determined by scintillation counting, and t equals time after onset of infusion.

Statistical Analyses

In each of 83 brain regions, effects on k* of NMDA and MK-801 alone and in combination in control diet rats were compared to effects of saline by one-way ANOVA with Dunnett's post-test with correction for four comparisons, using GraphPad Prism version 3.0a for Macintosh (GraphPad Software, San Diego CA, www.graphpad.com). Corrections for multiple comparisons across regions were not made because this was an exploratory study (Bland, 2000) to identify the regions that were involved in individual drug effects. Additionally, a two-way ANOVA, comparing Diet (LiCl vs control) with Drug (NMDA vs saline), was performed for each region using SPSS 10.0 for Macintosh (SPSS Inc., Chicago, IL, http://www.spss.com). At regions in which Diet × Drug interactions were statistically insignificant, probabilities of Main effects of Diet and Drug were separately calculated. At regions in which interactions were statistically significant, these probabilities were not calculated because they cannot be interpreted with certainty (Tabachnick and Fidell, 2001). Instead, unpaired t-tests were used to test for individual significant differences between means. Data are reported as means±SD, with statistical significance taken as p⩽0.05.

RESULTS

Physiological Parameters

At the time of surgery, the LiCl-fed animals weighed significantly less than the control diet group (257±42 vs 299±15 g, n=36, p<0.0001), as reported previously (Basselin et al, 2003).

In control diet rats, 25 mg/kg NMDA produced repeated cycles of head weaving and body movements lasting 4±1 s (activity period), following by a ‘calm period’ lasting 9±1 s (Table 1). The net time during which cycling continued was 90±15 s. NMDA at 50 mg/kg also caused cycling, but with a significantly shorter ‘calm period’ (5±2 s) than the lower dose, and a longer ‘net cycling period’ (180±10 s). No significant difference in cycling periods was observed in animals fed the LiCl vs those fed the control diet. MK-801 provoked head-weaving and body movements from 5 min after injection to the end of the experiment.

Table 1 Physiological Parameters and Behavior Activity Following Drug Administration in Unanesthetized Rats
Full size table

NMDA at both doses significantly decreased heart rate by about 20% in both control and LiCl diet groups, but did not significantly change arterial blood pressure, whereas MK-801 increased arterial blood pressure by 30% in the control diet group (Table 1). Such changes, reported previously, have been ascribed to a centrally mediated increase in sympathetic nerve activity by MK-801 (Lewis et al, 1989). MK-801 given before NMDA abolished NMDA's significant effect on heart rate.

Regional Brain AA Incorporation Coefficients, k*, in Control Diet Rats

Plasma input function

As shown by equation (1), values of k* for individual brain regions were calculated by dividing brain radioactivity by the brain exposure to plasma [1-14C]AA, the plasma input function equal to mean integrated arterial plasma radioactivity between 0 and 20 min. This function did not differ significantly between any groups. Mean integrated radioactivity, nCi/s/ml, for control diet rats: saline (159003±23885), NMDA 25 mg/kg (162630±23409), NMDA 50 mg/kg (155796±12953), MK-801 0.3 mg/kg (167647±16279), and MK-801+NMDA 50 mg/kg (179785±10361), and for the LiCl diet rats: saline (177699±14678), NMDA 25 mg/kg (166619±19830), NMDA 50 mg/kg (160236±15150). We have reported that chronic LiCl administration does not alter plasma unesterified fatty acid concentrations, including the AA concentration (Chang et al, 1999).

NMDA administration to control diet-fed rats

As illustrated in Figure 1, autoradiographs of coronal brain sections from control rats showed that NMDA 50 mg/kg compared with saline generally increased k* for AA whereas MK-801 generally decreased k*. There was no apparent difference between the autoradiographs from the LiCl-fed and control diet rat at baseline (i.p. saline), whereas NMDA in the LiCl-fed rat did not appear to have an effect on k*. Net regional data derived from the autoradiographs are presented in Tables 2 and 3 below.

Figure 1
figure 1

Effect of NMDA and MK-801 on AA incorporation into brain of control and LiCl fed rats. Regional incorporation coefficients k* are illustrated in autoradiographs from coronal brain sections obtained 20 min after infusing [1-14C]AA. Fr, frontal cortex; Mot, motor cortex; ACg, anterior cingulated cortex; Som, somatosensory cortex; CPu, caudate putamen; Hipp, hippocampus.

Full size image
Table 2 Mean Values of Arachidonic Acid Incorporation Coefficients k*a in Control Diet-Fed Rats, at Baseline (in Response to Saline) and in Response to NMDAb, MK-801c, and MK-801+NMDAd
Full size table
Table 3 Mean Values of Arachidonic Acid Incorporation Coefficients k*a in Lithium Diet-Fed Rats, at Baseline (in Response to Saline) and in Response to NMDAb
Full size table

As shown in Table 2, NMDA (25 mg/kg) compared with saline significantly increased k* for AA in 49/83 regions examined (25 regions at p<0.05 and 24 regions at p<0.001). k* was elevated significantly in the prefrontal, frontal, motor, somatosensory, auditory and visual cortical areas (18–38%), olfactory tubercle (22%), hippocampus (20–29%), caudate-putamen (17–20%), lateral geniculate nucleus dorsal (22%), regions of the thalamus (15–20%), and substantia nigra (20%), with mean increments of 21.8±5.1% in these areas. Also, 67 of 83 regions, including the nucleus accumbens (39%) (10 regions at p<0.05 and 57 regions at p<0.001), showed significant increments in k* following 50 mg/kg NMDA compared with saline. Consistent with a dose–response relation, the mean of the significant increments equaled 30.6±8.0%, which was significantly greater (p<0.0001) than that following 25 mg/kg NMDA.

MK-801 in control diet rats

Compared with saline, MK-801 given to control diet rats decreased k* for AA significantly in 21/83 regions (Table 2, Figure 1), including the frontal cortex (16%), pyriform cortex (20%), preoptic area (21%), suprachiasmatic nucleus (24%), globus pallidus (25%), hippocampus (21%), thalamus (22–26%), and hypothalamus (32–49%). The mean reduction in the significantly affected regions was 29.8±10.9%.

MK-801 followed by NMDA in control diet rats

In control diet rats, MK-801 followed by NMDA 50 mg/kg, when compared with saline decreased k* significantly in 21/83 regions, of which 17 showed a significant decrease following MK-801 alone (see above).

Insignificant Diet × Drug interactions: As illustrated in Table 3, 18/83 regions had a statistically insignificant Diet (LiCl diet vs control diet) × Drug (NMDA vs saline) interaction with regard to k* for AA*. In seven of the 18, NMDA compared with saline had a significant positive main Drug effect on k*, elevating k* to the same extent in both the control diet and LiCl-fed rats. In two of the 18 regions (preoptic area and thalamus posterior), LiCl compared with control diet had a significant main effect on k*, elevating k* to the same extent following saline or NMDA injection. Of the 83 regions, 10 did not have a significant Drug or Diet main effect.

Significant Diet × Drug interactions: Diet × Drug interactions were statistically significant in 65/83 regions, including the prefrontal, frontal, motor, pyriform, anterior cingulate, somatosensory and visual cortex, preoptic area, superchiasmatic nucleus, globus pallidus, hippocampus, caudate-putamen, habenular nucleus, lateral geniculate nucleus dorsal, geniculate medial, areas of the thalamus and hypothalamus, substantia nigra, and inferior colliculus (Table 3). In 10 of the 65 regions, unpaired t-tests showed that the LiCl diet, compared to the control diet, elevated k* significantly at baseline (in response to saline). These regions included visual cortex layers I, IV, and VI, lateral habenular nucleus, medial geniculate nucleus, and inferior colliculus, which belong to central auditory and visual pathways (Brodal, 1981). We reported this observation (Basselin et al, 2003, 2005b), and suggested that it may underlie lithium's ability in human subjects to increase the amplitudes of P1/N1 components of auditory-evoked responses and of 65-P95 and P95-N125 components of visual-evoked responses (Fenwick and Robertson, 1983; Hegerl et al, 1990; Ulrich et al, 1990).

In 65/83 regions, NMDA increased k* significantly in the control diet but not in the LiCl-fed rats. Thus, LiCl blocked 97% of the positive k* responses to 50 mg/kg NMDA (65/67).

DISCUSSION

This study shows that (1) the in vivo fatty acid method (see Introduction) (Rapoport, 2001, 2003) can be used to image PLA2 activation coupled to NMDARs in unanesthetized rats; (2) regional increments in k* for AA in response to subconvulsant NMDA doses in control diet rats are blocked by pretreatment with MK-801; (3) MK-801 alone (or when given prior to 50 mg/kg NMDA) produces widespread baseline reductions in k* for AA in control diet rats; and (4) a 6-week LiCl diet, sufficient to produce plasma and brain lithium concentrations therapeutically relevant to BD (Bosetti et al, 2002b; Chang et al, 1996), largely (97%) prevents the significant NMDA-induced increments in k* seen in control diet rats.

NMDA at 25 and 50 mg/kg significantly increased k* for AA by an average of 22% and 31%, respectively, in significantly affected brain regions. These regions, reported to have NMDARs (Bernard and Bolam, 1998; Ikeda et al, 2003; Jarvis et al, 1987; Monaghan and Cotman, 1986; Pal et al, 1999; Petralia et al, 1994), included cerebral cortex, caudate-putamen, globus pallidus, hippocampus, substantia nigra, and nucleus accumbens. NMDA has also been reported to increase the regional cerebral metabolic rate for glucose, rCMRglc, a marker of axonal activity (Sokoloff, 1999), in the rat suprachiasmatic nucleus (Shibata et al, 1992), frontal cortex layers V-VI, globus pallidus, caudate-putamen, substantia nigra, thalamus, nucleus accumbens, and lateral habenular nucleus (Browne et al, 1998).

The blocking of increments in k* for AA following NMDA by MK-801 is consistent with the increments being mediated by NMDARs. Furthermore, the ability of MK-801, given alone or prior to NMDA, to significantly reduce k* for AA by 16–49%, suggests that NMDA-initiated PLA2 activation contributes substantially to the baseline release of AA in control brain. This interpretation is consistent with evidence that glutamatergic receptors constitute about 75% of the synapses in the mammalian cortex, and that many of these synapses include NMDARs (Braitenberg and Schüz, 1998; Fonnum, 1984; Raichle and Gusnard, 2002).

The increments in k* for AA in response to the subconvulsant doses of NMDA were most likely due to enhanced Ca2+ entry into the cell via NMDARs, activating cPLA2 to release AA from membrane phospholipid. NMDA-provoked EPLA2 activation has been demonstrated in rat hippocampal slices, where it could be blocked by a specific cPLA2 inhibitor and was accompanied by the formation of lyso-phosphatidylcholine and glycerophosphocholine, consistent with activation of the AA-selective cPLA2 (Dennis, 1994; Weichel et al, 1999). Other studies also reported that Ca2+-mediated activation of PLA2 in neurons released AA following NMDA (Dumuis et al, 1988; Lazarewicz et al, 1990; Pellerin and Wolfe, 1991; Tapia-Arancibia et al, 1992). Although both cPLA2 and sPLA2 were activated in some of those studies (Kim et al, 1995), it is likely that mainly cPLA2 was activated in ours, as cPLA2 is sensitive to the low Ca2+ concentrations, 300 nM to 1 μM (Clark et al, 1995; Ismailov et al, 2004) that occur during normal neuronal activation. sPLA2 activation requires a much higher Ca2+ concentration, 16–18 mM, and is involved mainly in the presynaptic release of synaptic vesicles (Dennis, 1994; Matsuzawa et al, 1996; Wei et al, 2003). iPLA2 is Ca2+-independent and selective for DHA acid rather than AA (Dennis, 1994; Strokin et al, 2004).

The chronic LiCl diet of this study is reported to reduce global brain mRNA, protein and activity levels of cPLA2 in rats, but not to affect expression of sPLA2 or iPLA2 (Bosetti et al, 2002a, 2002b; Chang and Jones, 1998; Rapoport and Bosetti, 2002; Rintala et al, 1999; Weerasinghe et al, 2004). Its cPLA2 effect may be due a lithium-induced downregulation of Ca2+-dependent PKC, to reduce expression of PKC-dependent AP-2, which regulates cPLA2 transcription (Rao et al, 2005; Wang et al, 2001). Consistent with selective cPLA2 targeting, the LiCl diet reduces incorporation and turnover rates of AA but not of DHA in brain phospholipids of unanesthetized rats (Chang et al, 1996). Thus, one explanation for its suppression of the NMDA-induced increments in k* for AA in the present study is that it directly reduces the cPLA2 activation response to increased intracellular Ca2+.

Another possible explanation is that it changed the phosphorylation state of NMDAR subunits, preventing NMDAR-mediated Ca2+ entry into cells. For example, exposing cultured neurons to lithium, or injecting lithium into rats for 7 days, has been reported to reduce NMDAR-mediated Ca2+ influx by inhibiting NR2B tyrosine phosphorylation and Src tyrosine kinase (Hashimoto et al, 2002, 2003; Nonaka and Chuang, 1998), or by reducing NR2A tyrosine phosphorylation and interactions of the NR2A subunit with Src and Fyn mediated by PSD-95 in the rat hippocampus (Ma and Zhang, 2003; Ma et al, 2004). If blocking NMDAR-mediated Ca2+ entry into neurons were the primary reason for the reduced k* responses, then lithium's reduction of global cPLA2 expression in the rat brain (see above) could be secondary to a chronic downregulation of Ca2+ entry via defective NMDARs (Du et al, 2004; Gray et al, 2003; Karkanias and Papke, 1999). In any case, blocking NMDAR signaling via cPLA2 as contributing to lithium's therapeutic efficacy is consistent with evidence that other anti-BD drugs (Calabrese et al, 1995)—lamotrigine (Farber et al, 2002; McIntyre et al, 2004), carbamazepine (Hough et al, 1996; Kubota et al, 1994), and valproic acid (Zeise et al, 1991) have direct or indirect NMDAR-blocking properties.

Reduced NMDAR signaling as contributing to lithium's efficacy is consistent with evidence of disturbed NMDAR subunit expression in the BD brain. NR1 and NR2 genes are reported to confer susceptibility to BD (Itokawa et al, 2003; Mundo et al, 2003), and NMDAR density is decreased in the post-mortem BD hippocampus (Scarr et al, 2003). In other studies, NR2D mRNA was higher in the post-mortem BD striatum (Meador-Woodruff et al, 2001), NR1 mRNA was decreased in the BD dentate gyrus and CA3 of the hippocampus (Law and Deakin, 2001), and NR3A mRNA was decreased in the BD prefrontal cortex (Mueller and Meador-Woodruff, 2004).

It remains possible that the NMDA response involved neuroreceptors other than NMDA. For example, dopaminergic and glutamatergic/NMDA transmission are closely linked in the mammalian brain, particularly in the prefrontal cortex, striatum, and nucleus accumbens (Ferretti et al, 2005; Ikeda et al, 2003; Piomelli and Di Marzo, 1993; Rogue et al, 1990; Takeuchi et al, 2002; Tseng and O'Donnell, 2004; Wang et al, 2003). Medium-sized spiny neurons, the majority of neurons in the neostriatum, receive convergent glutamate-containing afferents from the neocortex and dopamine-containing afferents from the substantia nigra and ventral tegmentum (Murata et al, 2002). Dopamine receptor-linked signaling can affect the phosphorylation state and synaptic efficacy of NMDARs coexpressed on striatal neurons (Cepeda and Levine, 1998; Chase, 2004; Nair et al, 1998), whereas inhibition of NMDAR expression by antisense oligonucleotides can modulate pre- and postsynaptic expression of D2-like receptors in these neurons (Murata et al, 2002). Furthermore, both dopaminergic and glutamatergic pathways can modulate neuroplasticity involving cAMP, DARPP-32 (dopamine, cAMP-regulated phosphoprotein of 32 kDa), and ERK (extracellular signal-regulated kinase) (Gould and Manji, 2005; Valjent et al, 2005). In this regard, chronic lithium increased the level of DARPP-32 in the rat frontal cortex (Guitart and Nestler, 1992), which may explain its ability to activate the ERK-signaling cascade (Einat et al, 2003).

In view of the interplay between D2-like receptors and NMDARs, it is not surprising that chronic LiCl suppresses increments in k* for AA in unanesthetized rats given quinpirole (Basselin et al, 2005a), an agonist of D2-like receptors coupled to PLA2 via a G-protein (Bhattacharjee et al, 2005; Hayakawa et al, 2001; Huo and Healy, 1991; Vial and Piomelli, 1995). This suppression agrees with evidence that drugs that interfere with dopaminergic neurotransmission can be effective in BD (Bunney and Garland-Bunney, 1987; Bymaster and Felder, 2002; Fisher et al, 1991; Peet and Peters, 1995; Post et al, 1980; Sultzer and Cummings, 1989). On the other hand, the ability of chronic LiCl to potentiate k* responses in rats to the muscarinic M1,3,5 receptor agonist, arecoline (Basselin et al, 2003), is consistent with reports that cholinomimetics can be therapeutic as well (Bunney and Garland-Bunney, 1987; Bymaster and Felder, 2002). These published data, and the data of this study, suggest that lithium acts in BD by rectifying several neurotransmitter-signaling imbalances through downregulating NMDAR and D2-like receptor signaling and upregulating M1,3,5 receptor signaling involving PLA2 and the AA cascade (Bunney and Garland-Bunney, 1987; Bymaster and Felder, 2002; Dumuis et al, 1988; Rapoport and Bosetti, 2002; Shimizu and Wolfe, 1990).

Glutamate at excessive concentrations can kill neurons by the process of excitotoxicity (Coyle and Puttfarcken, 1993; Rothman and Olney, 1995), and it has been suggested that glutamatergic excitotoxicity contributes to BD (Zarate et al, 2003). Lithium's ability to block NMDA-initiated increments in k* for AA (this paper) is consistent with evidence that chronic lithium can be neuroprotective in animal models of neurodegeneration and even in BD patients (Moore et al, 2000; Rowe and Chuang, 2004; Zarate et al, 2003). Excess PLA2-mediated release of AA, with concurrent formation of excess lyso-phospholipids, can engender neurotoxic sequellae (Bazan et al, 1995; Farooqui et al, 1997; Sapirstein and Bonventre, 2000).

Finally, interference in the normal NMDAR-initiated PLA2-mediated release of AA by lithium may disturb long-term potentiation in the hippocampus, considered the basis of learning and memory (Dumuis et al, 1988; Nishizaki et al, 1999; Otmakhov et al, 2004; Schmitt et al, 2005; Williams et al, 1989). This may be relevant to reports that lithium treatment inhibited learning, memory, and speed of information processing in BD patients and to some extent in control subjects (Honig et al, 1999; Pachet and Wisniewski, 2003; Stip et al, 2000).