Update on the pathological roles of prostaglandin E2 in neurodegeneration in amyotrophic lateral sclerosis

The expressions of cPLA₂ and COX-2 are up-regulated in the spinal cords of ALS patients and ALS model mice [6, 42–44]. Nimesulide, a COX-2-selective inhibitor, suppresses PGE₂ elevation in the spinal cord, delays the onset of motor deficits, and tends to prolong the survival of G93A mice [45]. Furthermore, inhibition of COX-2 in G93A mice by celecoxib or rofecoxib, which are more selective than nimesulide, reduces the production of PGE₂, delays the onset of weight loss, ameliorates motor performance degeneration, extends survival, and protects against depletion of motor neurons in the spinal cord [26, 46]. We previously discovered that the protein level of mPGES-1 in the spinal cord was significantly increased after 15 weeks of age, which represents the early symptomatic stage, while the levels of cPGES and mPGES-2 remained constant with age [47]. Similar to the increased expression of mPGES-1, we found that the protein level of Iba1, a microglial marker, was also significantly increased in the G93A mice after 15 weeks of age. Surprisingly, our immunofluorescence staining assay demonstrated that mPGES-1 was co-localized with motor neurons both at 11 weeks (a pre-symptomatic stage) and at 15 weeks of age in these mice. On the contrary, co-localization of mPGES-1 with Iba1-positive microglia was detected in the spinal cords of G93A mice only at 15 weeks. Unlike the microglial marker Iba1, expression of the astrocytic marker glial fibrillary acidic protein (GFAP) was increased at 17 and 19 weeks of age (the symptomatic and end stage, respectively), with little co-localization with mPGES-1. These results suggest that, at the early symptomatic stage, mPGES-1, but not cPGES or mPGES-2, contributes to the PGE₂ increase in activated microglia in the spinal cord and to the progression of motor neuron degeneration in ALS.

Interestingly, AAD-2004, a dual-function drug derived from 5-aminosalicylate and sulfasalazine, has been reported to prevent PGE₂ production, microglial activation, and motor neuron degeneration through blockade of both mPGES-1 and free-radical production in G93A mice, resulting in amelioration of motor function and prolonged survival [48]. Moreover, the therapeutic effect of AAD-2004 is superior to that of ibuprofen, a type of non-steroidal anti-inflammatory drug (NSAID) [48]. While these results seem to support ours, it should be noted that the therapeutic effect of AAD-2004 is not only dependent on the inhibition of mPGES-1, but also requires direct free-radical scavenging. Deletion of mPGES-1 has been reported to suppress oxidative stress in angiotensin II-treated hyperlipidemic mice [49] and in 6-hydroxydopamine (6-OHDA)-treated mice [50]. These reports make it difficult to determine the degree to which free-radical scavenging contributes to the therapeutic efficacy of AAD-2004, and thus it is worth investigating the therapeutic effects of mPGES-1 inhibitors without direct free-radical scavenging activity in ALS, in comparison with the effects of combination of NSAIDs and antioxidants. In addition, AAD-2004 lowers PGE₂ levels without affecting the production of prostacyclin (PGI₂) in the lumbar spinal cord of G93A mice [48]. Considering that a sustained-release PGI₂ receptor agonist, ONO-1301-MS, improves motor function and prevents loss of motor neurons in G93A mice [51], the selective inhibition of PGE₂ production by mPGES-1 is more beneficial for the treatment of ALS than the non-selective inhibition of PG production by COX-2.

Our study described above revealed the involvement of synthetic steps of PGE₂ in ALS [47]. With regard to the degradation process of PGE₂, our previous study [27] detected a band immunoreactive for 15-PGDH in the spinal cord of G93A mice starting at 15 weeks (the early symptomatic stage), and its expression was significantly increased at 17 weeks and older. Immunohistochemical analysis showed that 15-PGDH was co-localized with GFAP-positive astrocytes at 19 weeks. Unlike mPGES-1, 15-PGDH did not co-localize with NeuN-positive motor neurons or Iba-1-positive microglia in the spinal cord of G93A mice. Surprisingly, despite an increase in 15-PGDH in astrocytes, PGE₂ levels in the lumbar spinal cord were increased in an age-dependent manner, with a significant increase at 19 weeks. These results suggest that the production of PGE₂ overwhelms the increase of 15-PGDH and its activity in astrocytes at the end stage of ALS. An in vitro study reported that the uptake of PGE₂ via the prostaglandin transporter, an organic anion transporter, as well as 15-PGDH, is required for PGE₂ catabolism [52]. Although further studies are required to determine the expression of the prostaglandin transporter in spinal astrocytes in G93A mice, these results suggest that astrocytes expressing 15-PGDH internalize PGE₂ synthesized by mPGES-1-expressing motor neurons and microglia to maintain PGE₂ homeostasis.

PGE₂ exerts its physiological functions by binding to the EPs. The activation of EP1 contributes to neurotoxicity in cultured rat mesencephalic neurons [55]. EP2 has been shown to induce caspase-3–dependent cell death in cultured rat cortical and hippocampal neurons [56, 57]. EP3 activation mediates the glutamate-induced excitotoxicity in cultured rat hippocampal slices [58]. In contrast, EP2 protects against glutamate cytotoxicity in cultured rat hippocampal neurons [59]. Furthermore, EP2 and EP3 have a protective effect against glutamate cytotoxicity induced by glutamate transporter inhibition in rat organotypic spinal cord slice cultures [60]. The EP4 receptor agonist ONO-AE1-329 attenuates injury in the brain in an acute murine model of the N-methyl-D-aspartate-induced excitotoxicity [61]. These studies suggest that PGE₂ exerts protective or pathogenic effects against neurotoxicity, depending on the subtype or the cell target of the activated EP. The NSC-34 cell line is the most studied motor-neuron cell line, and is produced by the fusion of mouse neuroblastoma cells with motor neurons taken from embryonic spinal cords [62]. Differentiated NSC-34 cells are reported to exhibit the unique morphological and physiological characteristics of primary motor neurons [62, 63]. Our group was the first to identify the distribution of EP subtypes and their effects on the PGE₂-induced cell death in differentiated NSC-34 cells [64]. EP2 and EP3 were highly expressed in mouse spinal cord motor neurons and in differentiated NSC-34 cells. Exposure of NSC-34 cells to exogenous PGE₂ and butaprost, an EP2-selective agonist, resulted in decreased cell viability in a concentration-dependent manner, whereas exposure to sulprostone, an EP1 and EP3 agonist, did not. These results suggest that activation of EP2, but not EP3, contributes to the PGE₂-induced motor-neuron death.

In support of our results [64], genetic deletion of EP2 has been reported to down-regulate the pro-inflammatory process, improve motor strength and extend survival of G93A mice [65], suggesting that the PGE₂–EP2 axis plays an important role in ALS neurodegeneration. Liang et al. also revealed that the expression of EP2 was increased in GFAP-positive astrocytes and Iba1-positive microglia of spinal cord [65]. We demonstrated that the EP2 immunoreactivity that co-localized with motor neurons in G93A mice at 15 weeks (the early symptomatic stage) was significantly more intense than that in age-matched wild-type mice and than that at 11 weeks (the pre-symptomatic stage) [66]. An in vitro study using differentiated NSC-34 cells showed that PGE₂ up-regulated the expression of EP2. The cells with an increased level of EP2 resulting from pre-treatment with PGE₂ were significantly more vulnerable to the PGE₂-induced cell death. Consistently, PGE₂ is involved in the up-regulation of EP1 and EP4 in dorsal root ganglion neurons from partial sciatic nerve-ligated rats [67] and in the up-regulation of EP1 and EP2 in primary rat hepatocytes [68]. Although the mechanism underlying the positive-feedback regulation of EP2 in vivo has yet to be clarified, these results suggest that PGE₂ increases the expression of EP2 in motor neurons during the early symptomatic stage of ALS, and that its increased expression directly drives neurodegeneration in ALS.

Mouse EP3 has three distinct splicing variants: the EP3α and the EP3β isoforms are coupled to G_αi and the EP3γ isoform is coupled to both G_αs and G_αi [80]. In our previous study, the expression of EP3α and EP3γ, but not EP3β, was detected in the spinal cords of wild-type and G93A mice [81], and the expression was not significantly different between wild-type and G93A mice. Next, laser-capture microdissection was performed to dissect out motor neurons from samples of lumbar spinal cord. In the motor neurons, EP3γ mRNA expression was predominant in both wild-type and G93A mice, whereas EP3α and EP3β mRNAs were not detected. Similar to the results in spinal cords, the expression level of EP3γ in motor neurons did not differ between G93A and wild-type mice [81]. Furthermore, EP3γ has predominant mRNA expression in differentiated NSC-34 cells [76]. Our previous studies showed that activation of EP3, unlike EP2, failed to contribute to PGE₂-induced cell death in motor neurons [64, 66, 76]. However, another study using rat organotypic spinal cord slice cultures showed that sulprostone had a protective effect against glutamate neurotoxicity [60]. Therefore, further investigations are warranted to determine the effect of EP3 on neuronal cell death induced by species other than PGE₂ and associated with the progression of ALS.

mPGES-1, 15-PGDH, and EP2 have been shown to be induced in the spinal cords in G93A mice. However, the underlying mechanism of the induction in ALS remains to be determined. One possible contributor to the induction of these expressions is pro-inflammatory cytokines. The spinal cords of mice injected with either IL-1β or TNF-α show increased expression of COX-2 [82]. TNF-α induces PGE₂ synthesis via increased expression of COX-2 and mPGES-1 in primary mixed spinal cord cells from mice [83]. Polyinosinic-polycytidylic acid increases PGE₂ production through induction of COX-2 and mPGES-1 synthesis in primary rat microglia [84]. In primary mouse astrocytes, PGE₂ expression can be induced by TNF-α [85] and IL-1β stimulation [86]. Although the mechanism of 15-PGDH expression in the nervous system is unknown, IL-4 has been reported to increase it in human lung cancer cells [87]. Furthermore, IL-1β increases the expression of EP2 and EP4 in primary rat hippocampal neurons [88], and EP3 expression in primary rat astrocytes and the U373 human astrocyte cell line [89]. However, in another study, IL-1β and TNF-α did not alter the mRNA expression of EP receptors in primary rat dorsal root ganglia cells [90], suggesting that the regulation of EP expression by proinflammatory cytokines depends on the cell type. These studies indicate that pro-inflammatory cytokines induce up-regulation of mPGES-1, 15-PGDH, and EPs under pathological conditions, and further studies are required.

Interestingly, however, PGE₂ suppresses the LPS-induced COX-2 and mPGES-1 expression and decreases TNF-α production via activation of EP2 in primary mouse mixed spinal cord cells [91]. Studies using the BV-2 mouse microglial cell line have also reported that PGE₂ suppresses the LPS-induced IL-18 expression [86] and the amyloid-β (Aβ)-induced TNF-α expression [92]. The latter may be mediated by EP4 [92]. These reports support for a protective role of PGE₂, contrary to our studies. Considering that COX-2-selective inhibitors [26, 45, 46] and a mPGES-1 inhibitor [48] suppress disease progression in G93A mice, the responsiveness to PGE₂ may differ between LPS-stimulated transient inflammatory conditions and the chronic inflammatory conditions of ALS. Most importantly, previous studies have reported that the primary astrocytes from G93A mice [85] and TDP-43–deficient microglia [93] show upregulation of COX-2 expression and PGE₂ production. G93A mouse-derived primary astrocytes have higher levels of basal TNF-α and elicit a greater increase in TNF-α expression after either a IFN-γ or a TNF-α challenge compared with those from non-transgenic mice [85]. Therefore, although further investigations are required, the chronic increase in basal levels of PGE₂ in SOD1^G93A mice may facilitate the transition of astrocytes into a more active inflammatory state. Furthermore, TDP-43-deficient microglia induced neuronal death in a co-culture system, which was suppressed by the COX-2 inhibitor celecoxib [93], suggesting that the glial cells with genetic variant associated with ALS, not limited to SOD1^G93A, contribute to neurodegeneration in ALS through elevated basal levels of PGE₂.

We here examine the role of the PGE₂-related enzymes and receptors in ALS and the cytotoxicity and interactions of PGE₂ relevant to motor neurons, and we discuss the importance of PGE₂ for the progression of ALS. This review has some limitations, but it also has some important implications for future studies.

First, even though COX-2 inhibitors improve motor function and prolong the life span of G93A mice [26, 45, 46], no anti-inflammatory drugs, including COX-2 inhibitors, have shown significant benefits in clinical studies in ALS patients [94]. The administration of celecoxib at 800 mg/day for 12 months did not improve muscle strength, vital capacity, estimated exercise units, or the ALS Functional Rating Scale-Revised, nor did it prolong survival [95]. The most notable issue is an absence of PGE₂ increase in the CSF of ALS patients, and that celecoxib did not reduce PGE₂ levels in the CSF of ALS patients [95]. However, other studies have reported increased levels of PGE₂ in the CSF of sporadic ALS patients [24, 25]. For instance, Cudkowicz et al. suggested that the failure to detect high levels of PGE₂ was due to the absence of a significant degree of inflammation at the disease stage studied, or due to the fact that PGE₂ levels in the CSF do not necessarily reflect COX-2 activity in the central nervous system [95]. This discrepancy may be explained by the expression and localization of mPGES-1 in the spinal cord [48]. Although the measured concentrations in clinical studies refer to levels in the CSF, the local tissue concentrations are more important because PGE₂ acts as an autocrine or paracrine factor in target cells. We observed that mPGES-1 co-localizes with motor neurons and activated microglia of the spinal cord [48], suggesting that the local up-regulation of mPGES-1 under pathological conditions contributes to increased local concentrations of PGE₂ and that PGE₂ can be produced locally even under COX-2 inhibition. Thus, inhibition of mPGES-1 or antagonism of EP2 remains useful therapeutic targets. However, the local cellular concentrations of PGE₂ in the spinal cord are unknown. Furthermore, no studies have tested mPGES-1 inhibitors [96] or EP2 antagonists in ALS patients, so their usefulness in humans is unclear; further studies are required. Most recently, it has been reported that the combination of ciprofloxacin and celecoxib restores the morphological defects and abnormal neuromuscular junctions of motor neurons and improves locomotor function in SOD1 G93R transgenic zebrafish and TDP-43-mutant zebrafish, compared to treatment with each drug alone [97]. Therefore, the potential synergistic effects of COX-2 inhibitors with other pharmaceuticals are of interest.

Second, the PGE₂–EP2–cAMP signaling pathway may be involved not only in motor-neuron cell death but also in motor-neuron differentiation [98]. We previously showed that PGE₂ induces morphological differentiation through EP2 activation in undifferentiated NSC-34 cells. Moreover, the cAMP signaling axis is at least partially involved in the molecular mechanisms of these effects [99]. Furthermore, it is noteworthy that PGE₂-differentiated cells have the physiological and electrophysiological properties of mature motor neurons [100]. Previous studies have reported that the COX-2 inhibitors meloxicam and nimesulide suppress neurogenesis in the olfactory bulb in the adult mouse brain [101], and that PGE₂ promotes the neural differentiation of mouse brain-derived neuroectodermal stem cells [102], suggesting that PGE₂ plays a key role in the differentiation of endogenous neural stem/progenitor cells into neural cells. Interestingly, neural progenitor cells were reported to migrate to regions near degenerated motor neurons in the spinal cord of G93A mice during the early symptomatic and symptomatic stages and then differentiate into neurons [103]. According to discussions of mPGES-1 in the first limitation, high concentrations of PGE₂ are present near degenerated motor neurons. Although further studies using adult neural progenitor cells derived from the spinal cord are required to clarify the role of PGE₂ in the differentiation of motor neurons, these shreds of evidence suggest that PGE₂ may also play an important role in a neurogenesis process that compensates for the neurodegeneration of ALS. Taken together, these results suggest new physiological roles for PGE₂, where it can exert multiple effects depending on the stage of neurodevelopment of spinal motor neurons. This supports the hypothesis that increased PGE₂ in the spinal cord, the mechanistic focus of ALS, is induced to promote the differentiation of new motor neurons, but the net effect is toxicity to surrounding mature motor neurons, resulting in the exacerbation of ALS.

The level of PGE₂ in the central nervous system is also increased in other neurodegenerative disorders, such as in the CSF in Alzheimer’s disease (AD) [104] and in the substantia nigra in Parkinson's disease (PD) [105].

AD is the most common cause of dementia/memory loss [106], characterized by accumulation of Aβ and neuronal tau proteins in the brain, which leads to the destruction of cholinergic neurons and damage to brain tissue. In the brains of AD patients, COX-1 accumulates mainly in microglia, whereas COX-2 accumulates mainly in neurons [107]. Furthermore, immunofluorescence staining of AD brain tissues showed that the expression of mPGES-1 was increased in neurons, microglia, astrocytes, and endothelial cells [108], and that mPGES-2 was markedly increased in pyramidal neurons [109]. Clinical trials have reported that treatment with indomethacin, a nonselective COX inhibitor, ameliorates cognitive impairment in patients with AD [110], and that naproxen, a selective COX-2 inhibitor, blocks the Aβ-mediated suppression of long-term potentiation and memory function in Tg2576 AD model mice [111]. Interestingly, PGE₂ induces the production of Aβ through activation of EP2 and EP4 both in vitro and in vivo [112, 113]. Genetic deletion of EP2 prevents oxidative brain damage, reduces Aβ levels, and improves the spatial memory of AD model mice [77, 114]. Unfortunately, as of May 27, 2023, there have been no clinical trials on AD that target the enzymes or receptors associated with PGE₂ [115].

Huntington’s disease (HD) is an autosomal dominant pathological disease characterized by a movement disorder and dementia [116]. Although the expression and distribution of PGE₂-related enzymes and receptors in HD are poorly understood, in vivo studies in R6/1 HD model mice showed that antagonization of EP1 by SC-51089 ameliorated memory and motor deficits [117], while activation of EP2 by misoprostol ameliorated long-term memory deficits through up-regulation of brain-derived neurotrophic factor in the hippocampus [118]. These reports suggest that EP could be a promising therapeutic target; to our knowledge, there has been no report on the clinical significance of PGE₂ in HD, so further investigation is warranted.

PD is characterized by abnormal α-synuclein aggregations known as Lewy bodies and Lewy neurites, resulting in bradykinesia due to massive death of the nigrostriatal dopaminergic neurons [119]. Consistent with increased PGE₂ [105], the expressions of COX-2 and mPGES-1 are up-regulated in the substantia nigra of post-mortem PD brains compared to those of healthy controls [50, 120]. Previous studies in the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine-induced mouse model of PD showed that DuP697, a selective COX-2 inhibitor, significantly reduced PGE₂ production and dopaminergic neurotoxicity [121], and that COX-2 deficiency combined with the COX-2-specific inhibitor valdecoxib attenuated the microglial activation, the loss of dopaminergic neurons, and the motor deficits [122]. However, a nested case–control study found no benefit of NSAIDs in PD [123]. Conversely, a meta-analysis by Gao et al. demonstrated that ibuprofen, unlike other NSAIDs or acetaminophen, lowers the risk of early PD progressing to PD compared to non-administration [124]; thus the neuroprotective effect of ibuprofen is noteworthy. In vitro studies reported that activation of EP1 was selectively neurotoxic to dopaminergic neurons in rat embryonic primary mesencephalic neuronal cultures [55], and that activation of EP2 by butaprost protected against the 6-OHDA-induced oxidative stress in midbrain-derived dopaminergic neurons from embryonic rats [79]. By contrast, ablation of EP2 enhanced microglia-mediated α-synuclein clearance in a co-culture of mouse microglia and mesocortical slices from patients with Lewy body disease. Ablation of EP2 also significantly reduced the level of neurotoxicity and α-synuclein aggregation in mice treated with 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine [125]. Because of these conflicting results [79, 125], the therapeutic effects of EP2 on PD warrant further investigation.

Multiple sclerosis (MS) is an autoimmune disease characterized by white-matter lesions attributable to loss of the myelin sheaths around the axons of neurons [126]. Although PGE₂ levels in MS patients are unknown, elevated expression of COX-1 and COX-2 and levels of PGE₂ have been observed in the cerebral cortex, cerebellum, and spinal cord of MS model mice with experimental autoimmune encephalomyelitis (EAE) [127]. In support of this, indomethacin, a non-selective COX inhibitor, attenuates the progression of EAE [128]. Furthermore, mPGES-1 upregulation occurs in macrophages in the spinal cord lesions of the EAE mouse model and in the brain tissues of MS patients [129]. mPGES-1-deficient mice have better clinical scores and suppressed responses of T helper 1 and helper 17 cells compared with non-deficient mice after EAE induction [129]. In EAE mice, PGE₂ has two opposing effects: it promotes the generation of T helper 1 and helper 17 cells through EP2 and EP4 while protecting the blood–brain-barrier via EP4 to prevent invasion of these cells into the brain [130].