Arachidonic Acid Metabolism in Brain Physiology and Pathology: Lessons from Genetically Altered Mouse Models

Abstract

The arachidonic acid (AA) cascade involves the release of AA from the membrane phospholipids by a phospholipase A₂ (PLA₂), followed by its subsequent metabolism to bioactive prostanoids by cyclooxygenases (COX) coupled with terminal synthases. Altered brain AA metabolism has been implicated in neurological, neurodegenerative and psychiatric disorders. The development of genetically altered mice lacking specific enzymes of the AA cascade has helped to elucidate the individual roles of these enzymes in brain physiology and pathology. The roles of AA and its metabolites in brain physiology, with a particular emphasis on the PLA₂ /COX pathway, are summarized, and the specific phenotypes of genetically altered mice relevant to brain physiology and neurotoxic models are discussed.

Background

The arachidonic acid (AA) cascade (Figure 1) is the biosynthetic pathway which involves the release of the n-6 polyunsaturated fatty acid AA from the sn-2 position of membrane phospholipids by a phosholipase A₂ (PLA₂) enzyme, and its subsequent metabolism to bioactive prostaglandins (PGs), thromboxanes, leukotrienes and epoxy fatty acids, by cyclooxygenase (COX), lipoxygenase (LOX) and cytochrome P450 epoxygenase enzymes, coupled to specific terminal synthases (Fitzpatrick and Soberman 2001). AA and its end-products are involved in physiological processes in the central nervous system such as synaptic signaling (Kaufmann et al. 1996), neuronal firing (Chapman and Dickenson 1992), neurotransmitter release (Ojeda et al. 1989), nociception (Hori et al. 1998), neuronal gene expression (Lerea and McNamara 1993; Lerea et al. 1997), cerebral blood flow (Pickard 1981; Li et al. 1997; Stefanovic et al. 2006), sleep/wake cycle (Hayaishi and Matsumura 1995), and appetite (Baile et al. 1973). Altered brain AA metabolism has been implicated also in a number of neurological, neurodegenerative, and psychiatric disorders, including epilepsy (McCown et al. 1997), ischemia (Sapirstein and Bonventre 2000b; Phillis and O'Regan 2003), stroke (Dirnagl et al. 1999), HIV-associated dementia (Griffin et al. 1994), amyotrophic lateral sclerosis (Almer et al. 2001), Alzheimer’s disease (Bazan et al. 2002), Parkinson’s disease (Teismann et al. 2003), schizophrenia (Muller et al. 2002), and mood disorders (Rapoport and Bosetti 2002; Sublette et al. 2004). Additionally, we have demonstrated that the release and/or metabolism of AA to PGE₂ via cytosolic PLA₂ (cPLA₂)/COX-2 in rat brain are common targets of therapeutic concentrations of mood-stabilizers used to treat bipolar disorder: lithium, valproic acid, and carbamazepine (Bosetti et al. 2002; Bosetti et al. 2003; Ghelardoni et al. 2004).

Figure 1. The arachidonic acid cascade.

The AA cascade involves the release of AA from membrane phospholipids by a PLA₂, followed by its metabolism to bioactive eicosanoids – prostaglandins, leukotrienes and epoxy fatty acids –through COX, lipoxygenase and cytochrome P450 epoxygenase enzymes, respectively.

EETs = epoxyeicosatrienoic acids; HETEs = hydroxyeicosatetraenoic acids; HPETE = hydroxyperoxyeicosatetraenoic acid; LOX = lipoxygenase; PG = prostaglandin; TXA₂ = thromboxane A₂

The regulation of brain AA metabolism and the coupling mechanisms between upstream and downstream enzymes involved in the prostaglandin biosynthetic pathway remain to be fully elucidated. While pharmacological inhibitors can be used to investigate the role of key enzymes involved in brain AA release and metabolism in physiological and pathological models, the lack, in some cases, of specific inhibitors or of a complete pharmacological inhibition, and standardization of dosing paradigms complicate the studies. In this regard, genetic mouse models lacking or over-expressing specific enzymes represent useful tools to identify the individual role of each enzyme and its coupling with the upstream and downstream enzymes in the AA cascade in physiology and pathology.

Genetic mouse models used to investigate the physiological and pathological roles of AA release and metabolism in brain

Cytosolic phospholipase A₂ alpha knockout (cPLA₂α ^−/−) mice

Among the members of the PLA₂ superfamily, cPLA₂α preferentially hydrolyzes arachidonoyl phospholipids and is activated by submicromolar concentrations of Ca²⁺ and by phosphorylation by mitogen-activated protein kinases (Leslie 1997) and protein kinase C-α (Li et al. 2007). The development of mice deficient in cPLA₂α (cPLA₂α ^−/−) by Bonventre’s and Shimizu’s groups (Bonventre et al. 1997; Uozumi et al. 1997) has helped to further elucidate the role of cPLA₂ in the central nervous system both under basal conditions and in pathological models.

Under basal conditions, although there was no difference in total phosphatidylcholine levels in the cPLA₂α ^−/− mice compared with wild type mice , the brain AA concentration of phosphatidylcholine was significantly reduced in cPLA₂α ^−/−mice (Rosenberger et al. 2003). We reported that cPLA₂α ^−/− mice show a marked decrease in the mRNA and protein levels of COX-2 as well as reduced COX activity, without any change in the expression of COX-1, 5-LOX, or cytochrome P450 epoxygenase, indicating that cPLA₂ and COX-2 are functionally coupled for PGE₂ biosynthesis (Bosetti and Weerasinghe 2003). Expression and activity of other PLA₂ enzymes was not significantly changed in cPLA₂α ^−/− mice, suggesting that type V sPLA₂ or type VI iPLA₂ do not compensate for the loss of cPLA₂ (Bosetti and Weerasinghe 2003). The role of cPLA₂ in the transcriptional regulation of COX-2 in brain was also highlighted by work from Sapirstein and colleagues (Sapirstein et al. 2005), who reported that after systemic administration of bacterial lipopolysaccharide (LPS), brain COX-2 mRNA levels are ~60% lower in cPLA₂α ^−/− mice, without any significant change in microsomal prostaglandin E synthase (mPGES)-1 mRNA induction. Brain levels of PGE₂ also were reduced after LPS injection in cPLA₂α ^−/− mice. Overall, the combined data indicate that cPLA₂α plays a significant role in regulating COX-2 gene expression and that COX-2 is the major contributor to brain PGE₂ formation.

cPLA₂α ⁻/⁻ mice also are less susceptible than wild type mice to cerebral ischemia/reperfusion injury after 2 h of middle cerebral occlusion (Bonventre et al. 1997; Tabuchi et al. 2003), with reduced neurological deficits, a reduction in infarct volume, and reduced brain edema compared to wild type mice (Bonventre et al. 1997; Tabuchi et al. 2003). The role of cPLA₂ in brain injury has been further investigated using the neurotoxicant 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), which damages substantia nigra neurons and causes a decrease in striatal dopamine concentration and is used as a model of Parkinson’s disease (recently reviewed in (Jakowec and Petzinger 2004)). cPLA₂α ^−/− mice show significantly less depletion of brain striatal dopamine after MPTP administration, suggesting that the increased resistance to MPTP neurotoxicity may be due to reduced release of free fatty acids and formation of lysophospholipids, as well as reduced production of reactive oxidative species via subsequent COX metabolism of AA (Klivenyi et al. 1998). Although the susceptibility of cPLA₂^−/− mice has not been investigated in other models of neurotoxicity, data currently available suggest that pharmacological inhibition of cPLA₂ may be promising in the treatment of neuroinflammation, ischemic brain damage, and possibly, Parkinson’s disease.

It should be taken into consideration that the mouse strains C57Bl/6n and 129/Sv have a naturally occurring missense mutation in the gene encoding for the type IIA sPLA₂. Consequently, cPLA₂α ^−/− mice, as well as their respective wild type mice, are deficient in type IIA sPLA₂ (Kennedy et al. 1995; Sapirstein and Bonventre 2000a). Although type IIA sPLA₂ is more abundantly expressed in the rat, whereas in the mouse the sPLA₂ isoform expressed at higher levels is type V sPLA₂ (Sawada et al. 1999), it would be worthwhile to identify how and to what extent sPLA₂ IIA genetic deletion contributes to the phenotype of cPLA₂α ^−/− mice. In this regard, transgenic mice overexpressing human sPLA₂-IIA are available (Grass et al. 1996) and could be used to identify the function of sPLA₂-IIA in the central nervous system in physiological and pathological conditions. In this regard, a recent report indicates up-regulated PLA₂-IIA mRNA brain expression and increased number of sPLA₂-IIA-immunoreactive astrocytes in the hippocampus and inferior temporal gyrus in Alzheimer's disease (AD) brains as compared to controls (Moses et al. 2006).

Cyclooxygenase-2 knockout mice

COX-2^−/− mice share some phenotypic similarities with cPLA₂^−/− mice, such as profound defects in female fertility (Morham et al. 1995; Bonventre et al. 1997), decreased susceptibility to ischemia/reperfusion brain injury (Bonventre et al. 1997; Sapirstein and Bonventre 2000b; Tabuchi et al. 2003), and MPTP-induced neurotoxicity (Klivenyi et al. 1998). These similarities between the two genotypes further support the evidence of a functional coupling between cPLA₂ and COX-2 enzymes (Naraba et al. 1998; Scott et al. 1999; Takano et al. 2000; Bosetti and Weerasinghe 2003).

Under basal conditions, the COX-2^−/− mouse brain shows altered brain lipid composition, including an increase in phosphatidyl serine and reduction in cholesterol (Ma et al. 2007), an upregulation of the activity and expression of the upstream Ca²⁺-dependent PLA₂ enzymes, cPLA₂ and sPLA₂, as well as a compensatory increase in the expression of the reciprocal isoenzyme, COX-1 (Bosetti et al. 2004). The increase in PLA₂ activity is consistent with a baseline elevation in AA incorporation coefficients (k*) in most brain regions, reflecting higher rates of AA loss in brain in COX-2^−/− mice compared to wild type mice (Basselin et al. 2006). Despite the compensatory increase in COX-1 expression, brain PGE₂ level is decreased by about 50% in the COX-2^−/−compared to wild type mice, further supporting the notion that COX-2 plays a major role in endogenous PGE₂ synthesis (Bosetti et al. 2004).

COX-2^−/− mice show a decrease in brain nuclear factor-κ B (NF-κ B) DNA-protein binding activity, likely caused by the decrease in the protein levels of p65, phosphorylated p65 and phosphorylated I-κ Bα (Rao et al. 2005). These changes in the NF-κ B transcription factor suggest that COX-2 deletion may in turn have profound transcriptional effects. Thus, we examined gene expression in the cerebral cortex and hippocampus of mice deficient in either COX-1 or COX-2. We found that COX-1^−/− or COX-2^−/− mice show significant changes in the expression of a multitude of genes in the hippocampus and cerebral cortex, with some occurring in both genotypes but most changes (>93–97%) being unique to one isoform deletion and not the other. The major function category affected by either COX-1 or COX-2 deficiency is transcriptional regulation, suggesting that COX end products can modulate gene expression. We also showed that the expression of genes constituting at least 5% of the total number of genes affected was involved in transmembrane and mitochondrial functions in the cerebral cortex of COX-2^−/− but not COX-1^−/− mice (Toscano et al. 2007).

Iadecola’s group demonstrated that constitutively expressed COX-2 plays a key role in the mechanisms coupling synaptic activity to neocortical blood flow. They show that selective COX-2 inhibition or genetic deletion of COX-2 attenuates the increase in cerebral blood flow produced by somatosensory activation. Vasodilator responses of the cerebral circulation to stimuli not dependant on neuronal activity, such as hypercapnia or topical application of acetylcholine or bradykinin, were not influenced by NS-398 and were preserved in COX-2^−/− mice (Niwa et al. 2000).

COX-2 seems to play a key role in the cholinergic signaling in brain as the AA incorporation response to arecoline, evident in wild type mice at sites of muscarinic receptors, is entirely absent in COX-2^−/− mice, suggesting that COX-2 is required for muscarinic receptor activation (Basselin et al. 2006). This implies that the response to arecoline mediated by M_1,3,5 receptors coupled to PLA₂ largely reflects the conversion of released AA to prostanoids by COX-2 (Basselin et al. 2006).

Because COX-2 is thought to be implicated in a number of neurological and neurodegenerative diseases, COX-2^−/− mice also have been used to investigate the role of COX-2 in different neurotoxic models. Similarly to cPLA₂^−/− mice, COX-2^−/− mice also are more resistant to MPTP-mediated neurotoxicity, showing a reduced loss of tyrosine hydroxylase-immunoreactive neurons in the substantia nigra pars compacta compared to wild type mice (Feng et al. 2002; Feng et al. 2003).

Iadecola and colleagues showed that brain injury produced by middle cerebral artery occlusion or following NMDA microinjection in the neocortex is reduced in COX-2^−/− compared to wild type mice (Iadecola et al. 2001b). They further reported that the lesion size produced by direct injection of NMDA in the somatosensory cortex is attenuated in COX-2^−/− mice and they determined, using COX-2^−/− mice and the COX-2 selective inhibitor NS-398, that COX-2 is not a major source of oxygen radicals after cerebral ischemia, suggesting that other COX-2 reaction products, including prostanoids or non-oxygen-based radicals may be responsible for brain injury (Kunz et al. 2006). In a model of post-ischemic brain injury after bilateral common carotid artery occlusion, COX-2^−/− mice also have reduced ischemic neuronal damage, measured by cresyl violet and TUNEL staining, in the hippocampus (Sasaki et al. 2004).

The role of PGE₂ or in general of COX-2 mediated prostaglandin metabolism in mediating or attenuating glutamate excitotoxicity and in kainic acid-mediated excitotoxic brain damage remains controversial. Several studies indicate that PGE₂ enhances glutamate release and excitatory amino acid-induced synaptic depolarization (Kimura et al. 1985; Sekiyama et al. 1995; Bezzi et al. 1998; Sanzgiri et al. 1999; Katafuchi et al. 2005). Furthermore, a recent report suggests that COX-2^−/− mice may have decreased PGE₂ release and reduced neuronal loss after direct microinjection of kainic acid in the CA3 region of the hippocampus (Takemiya et al. 2006). However, since that study did not report neuronal count under basal conditions, it cannot be excluded that COX-2^−/− mice show decreased neuronal count independently of kainic acid injection. On the other hand, other studies suggest that PGE₂ is protective against glutamate toxicity (Akaike et al. 1994; Cazevieille et al. 1994) and data from our lab indicate that when kainate is administered systemically by intraperitoneal injection, COX-2^−/− mice exhibit significantly higher seizure intensities in the 2 h following injection. Increased seizure intensity is accompanied by increased neuronal damage 24 h and after injection of kainate (Toscano and Bosetti, unpublished data).

Data from our group also suggest that COX-2^−/− mice show a higher neuroinflammatory response than wild type mice after acute intracerebroventricular injection of LPS, with increased activation of microglia and astrocytes and increased expression of pro-inflammatory cytokines, such as interleukin (IL)-1β and tumor necrosis factor (TNF)-α , and PGE₂ (Aid et al., unpublished data). These data agree with evidence that the COX-2 selective inhibitor NS-398 significantly increases the ability of LPS to trigger an innate immune response (Blais et al. 2005).

COX-2 overexpressing mice

Two independent transgenic mouse models with neuronal overexpression of the human COX-2 developed by Pasinetti’s (Kelley et al. 1999) and Andreasson’s groups (Andreasson et al. 2001) have been used to investigate how COX-2 modulates the susceptibility to brain injury. COX-2 transgenic mice show increased lethality following intracerebroventricular injection of kainic acid (Kelley et al. 1999), suggesting that COX-2 may be detrimental to neurons after direct injection of kainate. This observation may be related to a difference in prostaglandin profile in COX-2 overexpressing mice, or to the neuronal localization of overexpressed COX-2. Transgenic mice overexpressing functional COX-2 in neurons of the hippocampus, cortex, and amygdala also develop age-dependent cognitive deficits that are associated with a parallel age-dependent increase in glial fibrillary acidic protein and neuronal apoptosis (Andreasson et al. 2001). Opposite to the reduced susceptibility to ischemic brain injury of COX-2^−/− mice, a significant increase in infarct size was observed after middle cerebral artery occlusion in COX-2 transgenic mice, suggesting that chronic increases in COX-2 expression and enzymatic activity, which can occur in pathological states characterized by oxidative stress and chronic inflammation, could negatively affect neuronal survival in cerebrovascular disease (Dore et al. 2003).

COX-1 knockout mice

Like COX-2^−/− mice, COX-1^−/− mice have up-regulated brain expression and activity of Ca²⁺-dependent cPLA₂ and sPLA₂, as well as an increase in the reciprocal isozyme, COX-2 (Choi et al. 2006). In contrast to COX-2^−/− mice, COX-1^−/− mice have increased brain levels of PGE₂ and increased activation of NF-κ B, which involves the up-regulation of protein expression of the p50 and p65 subunits of NF-κ B, as well as increased protein levels of phosphorylated Iκ Bα and of phosphorylated IKK (Iκ B kinase) α/β (Choi et al. 2006), suggesting that the two isoforms have unique and distinct functions in brain AA metabolism, and that a deficiency of one cannot be compensated for by overexpression of the other. Microarray analysis identified 94 genes differentially expressed in the cerebral cortex and 182 genes differentially expressed in the hippocampus of COX-1^−/− mice compared to wild type mice. The expression of genes belonging to the functional group of kinases (> 5% of total number of genes) was selectively changed in the hippocampus of COX-1^−/− but not of COX-2^−/− mice (Toscano et al. 2007). In both COX-1^−/− and COX-2^−/− mice, the greatest number of genes whose expression was significantly changed in both brain regions belonged to the function category of transcriptional regulation (Toscano et al. 2007). Although some of the molecular mechanisms underlying these changes are not understood at this time, these data indicate that the two COX- isoforms have some independent effects on modulating gene expression.

COX-1 plays an important role in the maintenance of resting cerebral blood flow and in the vasodilation produced by selected endothelium-dependent vasodilators or hypercapnia (Niwa et al. 2001). Indeed, COX-1^−/− mice show up to 20% decrease in resting cerebral blood flow in selected telencephalic regions and attenuated responses to hypercapnia, bradykinin, AA, and the calcium ionophore A23187, whereas responses to acetylcholine, vibrissal stimulation, and S-nitroso-N-acetylpenicillamine are not affected (Niwa et al. 2001).

The differences between COX-1^−/− and COX-2^−/− mice are even more dramatic in the response to neurotoxic insults (Table 1). Whereas COX-2^−/− mice show increased protection against ischemic brain injury or and MPTP-induced neurotoxicity (Iadecola et al. 2001b; Feng et al. 2002; Feng et al. 2003), COX-1^−/− mice either show no difference compared to wild type mice (Cheung et al. 2002; Feng et al. 2003) or, in case of ischemia-reperfusion, they seem to be even more susceptible (Iadecola et al. 2001a).

Table 1.

Phenotypes of cPLA₂-, COX-1- or COX-2-deficient mice under basal conditions or upon different neurotoxic injuries.

BASAL	cPLA2−/−	COX-1−/−	COX-2−/−
cPLA2 expression	n.d.(Bonventre et al. 1997; Bosetti and Weerasinghe 2003)	↑ (Choi et al. 2006)	↑ (Bosetti et al. 2004)
sPLA2 expression	↔(Bosetti and Weerasinghe 2003)	↑ (Choi et al. 2006)	↑ (Bosetti et al. 2004)
COX-1 expression	↔ (Bosetti and Weerasinghe 2003)	n.d.(Choi et al. 2006)	↑ (Bosetti et al. 2004)
COX-2 expression	↓ (Bosetti and Weerasinghe 2003)	↑ (Choi et al. 2006)	n.d. (Bosetti et al. 2004).
NF-κB activity	unknown	↑ (Choi et al. 2006)	↓ (Rao et al. 2005)
PGE2 levels	↔ (basal)(Sapirstein et al. 2005) ↓ (after LPS)(Sapirstein et al. 2005)	↑ (Choi et al. 2006)	↓ (Bosetti et al. 2004)
SUSCEPTIBILITY TO NEUROTOXIC INJURY
Ischemia/Reperfusion	↓ (Bonventre et al. 1997; Sapirstein and Bonventre 2000b; Tabuchi et al. 2003)	↑ (Iadecola et al. 2001a) ↔ (Cheung et al. 2002)	↓ (Iadecola et al. 2001b)
MPTP-neurotoxicity	↓ (Klivenyi et al. 1998)	↔ (Feng et al. 2002)	↓ (Feng et al. 2003)
LPS- neuroinflammation	↓ (Sapirstein et al. 2005)	↓ (Choi et al. unpublished)	↑ (Aid et al., unpublished)
NMDA-icv induced cortical lesion	unknown	unknown	↓ (Iadecola et al. 2001b)
Kainic acid-induced seizures	unknown	↔ (Toscano et al., unpublished)	↑ (Toscano et al., unpublished)
Kainic acid-induced neuronal damage	unknown	↔ (Toscano et al., unpublished)	↑ ip (Toscano et al., unpublished) ↓ icv (Takemiya et al. 2006)

Although the susceptibility of COX-1^−/− mice to kainate-induced seizures and neuronal damage needs to be fully elucidated, preliminary results from our group indicate that the susceptibility of COX-1^−/− mice to 10 mg/kg of kainate, a dose that causes seizures and hippocampal neuronal damage in the COX-2^−/− mice, is not significantly different compared to wild type mice (Toscano and Bosetti, unpublished data). While COX-1^−/− mice do not show increase resistance to most of the brain insults examined, data from our lab indicate that COX-1^−/− mice have a reduced neuroinflammatory response and protein oxidative damage after acute intracerebroventricular injection of LPS (Choi et al., unpublished data). Therefore, it is very important to elucidate the specific roles that COX-1 and COX-2 play in the neuroinflammatory cascade of events that lead to brain injury.

mPGES-1 knockout mice

The contribution of mPGES-1 to brain PGE₂ production in response to LPS has been examined using mPGES-1^−/− mice (Ikeda-Matsuo et al. 2005). Following the injection of LPS into the substantia nigra, which contains a high density of microglia, mPGES-1 expression is specifically induced in activated microglia at the site of the localized inflammatory response. mPGES-1 deficiency completely abolished LPS-stimulated PGE₂ production in microglial cells (Ikeda-Matsuo et al. 2005). Constitutive expression of mPGES-2, cytosolic PGES (cPGES) and COX-1 is not significantly different in mPGES-1^−/− derived and wild-type derived microglia, indicating that mPGES-1 deficiency does not affect the expression of other enzymes involved in PGE₂ biosynthesis (Ikeda-Matsuo et al. 2005).

PGD₂ receptor (DP) knockout mice

Using both PGDS^−/− and DP^−/− mice, Mohri and colleagues demonstrated that PGD₂ plays an important role in microglia/astrocyte interaction and is involved in the pathological response to demyelination in twitcher mice, a model of human Krabbe's disease (Mohri et al. 2006). Specifically, PGDS^−/− or DP₁^−/− twitcher mice showed a remarkable suppression of astrogliosis and demyelination, as well as a reduction in twitching and spasticity.

The PGD₂/DP axis also has an important physiological role in non-rapid eye movement (REM) sleep, as demonstrated by the observation that infusion of PGD2 into the lateral ventricle of the brain increased the amount of non-REM sleep in wild type but not in DP^−/− mice (Mizoguchi et al. 2001).

PGE₂ receptors (EP1-4) knockout mice

The development of mice lacking the different subtypes of the G-protein-coupled receptors for PGE₂ - EP1, EP2, EP3, and EP4 - (Ushikubi et al. 1998) helped to further understand the physiological or pathological roles of these receptors. A number of studies using EP1^−/− mice indicate that the prostaglandin EP1 receptor mediates neurotoxicity after NMDA-induced excitotoxicity, oxygen glucose deprivation, or focal ischemia (Ahmad et al. 2006; Kawano et al. 2006). EP1^−/− mice also demonstrate behavioral abnormalities, including increased impulsive aggression, reduced social interaction, enhanced acoustic startle response, and jumping behavior in the cliff-avoidance test under conditions of social or environmental stress, suggesting that EP1 selective agonists might be used to treat impulsive behavior (Matsuoka et al. 2005). EP1^−/− and EP3^−/− mice also show impaired HPA activation after i.p. LPS injection, with a significant decrease in plasma levels of ACTH compared to wild type mice (Matsuoka et al. 2003).

The role of the EP2 receptor as elucidated with EP2^−/− mice appears to be more complex. Microglial EP2 has been shown to be critical to neuronal damage from activation of cerebral innate immunity involving COX-2 and inducible nitric oxide synthase (Montine et al. 2002; Shie et al. 2005). On the other hand, the EP2 receptor has been shown to mediate neuroprotection in an organotypic hippocampal model of NMDA-mediated excitotoxicity and in permanent focal ischemia induced by middle cerebral artery occlusion (Liu et al. 2005). The EP2 receptor mediates protection also in a chronic model of excitotoxic motor neuron degeneration (Bilak et al. 2004), as well as in β -amyloid neurotoxicity (Yagami et al. 2003) and neuroinflammation (Kim et al. 2002; Lee et al. 2004).

EP3 receptors, but not the other subtypes of EP receptors, play important roles in mediating the febrile response after PGE₂, IL-1β , or LPS (Ushikubi et al. 1998), by generating hyperthermia (Lazarus 2006). EP3 receptors are also involved in the ACTH response to bacterial LPS (Matsuoka et al. 2003).

Stimulation of EP4 receptors has been shown to protect cultured neurons against oxidative stress and cell death induced by β -amyloid exposure (Echeverria et al. 2005). In opposition to EP3 receptors, EP4 responses generate hypothermia (Lazarus 2006) and have been reported to activate the histaminergic system to mediate PGE₂- induced wakefulness (Huang et al. 2003). Since most EP4^−/− mice die during the neonatal period (Nguyen et al. 1997; Segi et al. 1998), data on the effect of EP4 receptor deletion in brain physiological or pathological conditions are lacking.

LOX knockout mice

ApoE^−/− mice crossbred with 12/15-LOX^−/− mice have significantly reduced markers of lipid peroxidation and of protein oxidation when compared with 12/15-LOX-expressing controls, indicating that 12/15-LOX modulates oxidative reactions in the central nervous system (Chinnici et al. 2005). Consistent with this observation, 12/15-LOX^−/−mice appear to be protected against neuronal cell death and oxidative stress caused by transient focal ischemia (van Leyen et al. 2006). Additionally, neurons isolated from 12-LOX^−/− mice are resistant to glutamate-induced death, suggesting an involvement of 12-LOX in glutamate-induced neurodegeneration (Khanna et al. 2003).

On the other hand, the involvement of 5-LOX in ischemic brain injury remains to be clarified. While Kitagawa and colleagues showed that 5-LOX^−/− and wild type mice do not differ significantly in brain infarct size following permanent and transient 60 min ischemia, suggesting that 5-LOX-derived leukotrienes are not involved in infarct expansion in focal cerebral ischemia (Kitagawa et al. 2004), a recent report indicates that 5-LOX inhibitors protect brain in a rat model of focal cerebral ischemic injury (Jatana et al. 2006). 5-LOX^−/− mice show increased resistance to traumatic spinal cord injury by showing a marked decrease in neutrophils infiltration, cytokine expression apoptosis, and histological damage and ameliorated motor recovery as compared to wild-type mice (Genovese et al. 2005). 5-LOX may also be implicated in affective behaviors, as 5-LOX^−/−mice show reduced anxiety-like behavior in various behavioral tests (Uz et al. 2002).

Conclusions

The use of genetically altered mouse models lacking specific enzymes of the AA cascade has helped to identify the metabolic coupling of these enzymes and their distinct roles in brain physiological conditions as well as in models of neurotoxicity. A summary of the different phenotypes under basal conditions and after different neurotoxic challenges is shown in Table 1. The data obtained under basal conditions indicate that cPLA₂ and COX-2 are functionally coupled in the synthesis of endogenous brain PGE₂ (Bosetti and Weerasinghe 2003; Bosetti et al. 2004).

Both cPLA₂^−/− and COX-2^−/− mice exhibit decreased susceptibility to ischemic brain injury (Bonventre et al. 1997; Iadecola et al. 2001b) and MPTP-induced neurotoxicity (Klivenyi et al. 1998; Feng et al. 2003). However, our data with acute LPS-induced neuroinflammation and kainic-acid-induced seizures and neuronal damage indicate that COX-2^−/− mice are more vulnerable than wild-type mice (Aid et al., unpublished data; Toscano and Bosetti, unpublished data). Therefore, it is quite possible that the increased or decreased susceptibility to neurotoxicity of COX-2^−/− mice is stimulus-specific. While COX-1^−/− mice do not appear to respond differently to most of the neurotoxic stimuli investigated (Cheung et al. 2002; Feng et al. 2002); Toscano and Bosetti, unpublished data), their decreased neuroinflammatory response to acute LPS (Choi et al., unpublished data) suggests that COX-1 plays a role in initiating the neuroinflammatory cascade, likely because of its preferential localization in microglial cells (Yermakova et al. 1999; Ballou 2000; Schwab et al. 2002). Not only do these observations demonstrate that COX-1 and COX-2 have unique roles in brain physiology, but being that neuroinflammation is a key component of some neurodegenerative diseases, including Alzheimer’s disease, our findings may have therapeutic implications. As recently noted by McGeer and Mc Geer, reactive microglia expressing COX-1 is the proper target for inflammatory suppression (Hoozemans et al. 2002), suggesting that non steroidal anti-inflammatory drugs (NSAIDs) with predominant COX-1 rather than COX-2 inhibition may be more effective in reducing brain inflammation (McGeer and McGeer 2006).

In summary, although data obtained in genetically altered mouse models should be interpreted with care because of possible compensatory changes, knockout and transgenic mice provide a useful approach to identify the roles of specific enzymes of the AA cascade in physiological and pathological models. Therefore, these genetic animal models combined with conditional knockouts, models of in vivo RNA interference, and the use of specific pharmacological inhibitors, may help to develop novel therapeutic strategies for diseases involving altered brain AA metabolism.