1. Introduction
Polyamines (PAs) are organic polycations found ubiquitously in organisms, and, in mammals, they are mainly represented by putrescine (Put), spermidine (Spd), spermine (Spm) and their acetylated forms. In mammalian cells, PAs are involved in cell proliferation, differentiation, apoptosis, the synthesis of proteins and nucleic acids, the regulation of ion channel activity and the protection from oxidative injury
[1][2][3]. Animals devoid of PA biosynthesis do not survive the early stages of embryonic development; notably, the supplementation of agmatine, a compound that belongs to the PA family and may serve as a precursor for Put (although exerting largely separate functions in mammal tissues), may be sufficient to rescue PA biosynthesis when the biosynthesis of Put is blocked (suggesting that the agmatine pathway is fully developed only later in life; for a discussion, see
[4]). In fact, agmatine is a promising candidate for the treatment of several disorders, including neurodegenerative diseases, and highlights key roles for PAs in central nervous system (CNS) disorders
[5].The cellular content of PAs is tightly regulated
[1][2][3]; their biosynthesis is catalyzed by different enzymes including S-adenosylmethionine decarboxylase (AdoMetDC), ornithine decarboxylase (ODC), spermine synthase (SMS) and spermidine synthase (SRM)
[6], while the enzymes N1-acetylpolyamine oxidase (PAOX), spermidine/spermine N1-acetyltransferase (SAT1) and spermine oxidase (SMOX) are involved in the PA catabolism (
Figure 1)
[7][8][9][10].
Figure 1. Enzymes involved in PA biosynthesis (encircled) and catabolism (boxed). ODC, ornithine decarboxylase enzyme; PAOX, N1-acetylpolyamine oxidase; SAT1, spermidine/spermine N1-acetyltransferase; SMS, spermine synthase; SRM, spermidine synthase. The enzyme spermine oxidase (SMOX), which is overexpressed in the Dach-SMOX mice model, is highlighted in red.
2. Polyamines in the Brain
Polyamines present a unique biochemistry in the brain, being primarily synthetized in neurons, while glial and microglial cells are mainly involved in their uptake and release. Furthermore, glial and microglial cells convert and oxidize PAs and release, besides hydrogen peroxide (H
2O
2), 3-aminopropanal (3-AP) and 3-acetoamidopropanal, hypusine, putreanine and gamma-aminobutyric acid (GABA) as end-products as well
[4][11]. The major pathway of PAs to enter the brain was found through transporters on the astrocyte endfeet enwrapping blood vessels at the glial–blood interface
[12], turning the focus on the mechanisms of uptake and replenishment in the case of an age-dependent decrease in PA synthesis. The most relevant PA transport system includes large pores such as connexins and pannexin hemichannels and transporters such as polyspecific organic cation transporters (OCTs), including solute carrier (SLC) 22A1-3
[4]. Such systems also transport monoamines such as dopamine or levodopa (
l-dopa)
[13] and may function in reverse mode releasing PAs, therefore regulating the extracellular PA levels and neuronal activity
[12]. Another important transporter for PAs is the vesicular transporter SLC18B1, present in both neurons and astrocytes
[14], which was found to be involved in regulating the PA content, the function of the GABA and glutamatergic systems, memory, synaptic function and plasticity
[15][16]. Recently, the P-type ATPase transporter ATP13A3 has also been demonstrated to be involved in PA mammalian import
[17][18]. In the adult brain, neurons are capable of synthetizing PAs
[4][19][20], and glial cells but not neurons accumulate PAs
[4][21][22][23][24]. However, the work of Masuko et al.
[25] reported that PAs, especially Spm, are accumulated in synaptic vesicles and released by depolarization.
The fact that PAs such as Spd and Spm are taken up and accumulated by glial cells has led to a major focus on astrocytes. Notably, astrocytes, regardless of the true glia-to-neuron ratio and whether they outnumber neurons in the human brain or not
[26][27], are no longer the “unacknowledged partner”
[28] but are now fully “acknowledged” partners of neurons in CNS
[29][30]. Two main functional consequences of PAs being stored in astrocytes can be envisaged: (a) PAs released from glia regulate the function of receptors and channels
[31][32] in glia and neurons; (b) PAs stored in astrocytes regulate their own glial inward rectifier K
+ (Kir)4.1 channels
[33][34], connexin-43 (Cx43) channels
[35][36] and GluA2-lacking alpha-amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid (AMPA) receptors
[37]. In fact, PAs released from astrocytes can affect both neurons and astrocytes while intracellular PAs regulate glial function. Notably, the astrocytic PA-sensitive receptors and channels, as well as the Put-to-GABA conversion
[38], play roles in CNS diseases related to PAs
[31][39]. On the other hand, astrocytes, by accumulating PAs in CNS
[14][22][23], are involved in the protection against disease(s) and may be relevant to the Spd therapeutic potential
[40][41][42][43][44][45][46].
Increasing evidence indicates that the dysregulation of the PA system is involved in neurodegeneration in different CNS pathological conditions. In the Snyder–Robinson syndrome, a mutation of the
SMS gene, leading to a reduction of Spm and an accumulation of Spd, is responsible for a complex syndrome with intellectual disability, movement disorders and seizures
[47]. Neurological abnormalities have been also reported in rodent models of the altered synthesis and catabolism of PAs
[10][48]. Altered cellular levels of PAs and PA dyshomeostasis have been implicated in numerous brain diseases, including mental disorders, epilepsy
[49], Alzheimer’s disease (AD)
[50][51], Parkinson’s disease (PD)
[52], traumatic brain injury
[53] and in the pathogenesis of ischemic brain damage
[54][55] and neurovascular damage in the retina
[56][57]. In most of these conditions, dysregulation of the enzyme SMOX was reported. SMOX activity was also found to be increased in sera from schizophrenia patients, suggesting a key role of SMOX in the pathology
[58]. Interestingly, in patients with neurocognitive impairment, SMOX over-expression caused by
SMG9 Nonsense Mediated MRNA Decay Factor (
SMG9) loss-of-function, a gene key regulator of nonsense-mediated decay, was associated with intellectual disability
[59].
3. Polyamine Catabolism: The Enzyme SMOX
Spermine oxidase catalyzes the conversions of Spm to Spd with the production of H
2O
2 and 3-AP
[60][61][62] (
Figure 2).
Figure 2. Spermine oxidase chemical reaction. Spermine (Spm) is oxidized to produce spermidine (Spd), 3-aminopropanal (3-AP) and hydrogen peroxide (H
2O
2).
Spermine oxidase is a highly inducible enzyme
[57] which is expressed mainly in the brain and skeletal muscle but also in the kidney, pancreas, bone marrow, lung, heart, intestine and spleen
[10][63], where it regulates the Spm/Spd ratio to balance the cellular PA content, while SMOX dysregulation can alter the PA homeostasis
[10][64]. The SMOX substrate Spm plays important functions in brain, since intracellular Spm acts as a neuromodulator responsible for the rectification of strong inward rectifier Kir channels
[25][65][66] of AMPA and of kainate Ca
2+-permeable receptors
[67][68]. On the other hand, extracellular Spm can affect the function of
N-methyl-D-aspartate (NMDA) glutamate receptors
[25][65][66]. Additionally, Spd, the oxidation product of SMOX, functions as a neuromodulator, even though it is less potent than Spm
[65]. New roles for Spd are increasingly investigated, such as its potential in protecting organisms from age-induced memory impairment through an autophagy-dependent homeostatic regulation at synapses
[40] or in enhancing cerebral mitochondrial function and cognition in aging
[42]. In fact, Spd has been reported to induce autophagy in model systems including rodent tissues and cultured human cells
[43][69][70]. In
D. melanogaster, autophagy seemed crucial for Spd protection against the presynaptic active zone changes during aging
[41]. SMOX activity, in addition to keeping PA cellular content balanced, can alter the cellular redox homeostasis by producing H
2O
2, an endogenous reactive oxygen species (ROS). Although H
2O
2 plays roles in physiological brain function, excessive H
2O
2 production can result in learning and memory impairment
[71]. Moreover, SMOX activity may be responsible for secondary tissue damage due to the generation of 3-AP, which spontaneously converts into acrolein
[9][72], further inducing inflammation and apoptotic cell death in an injured brain
[72].
Remarkably, growing evidence indicates a link between PA dyshomeostasis and neurodegenerative diseases. Polyamines are pivotal players in signalling responses to various environmental stimuli, which are involved in various aspects of the cellular metabolism, the maintenance of antioxidant capacity and osmotic regulation
[1]. It is becoming clear that the PA role may shift from positive to negative in disease development, contributing to the shift from healthy to pathological conditions. The regulation of the PA system entails the balance of PA neuroprotective effects (in healthy conditions) versus detrimental effects of PA derivatives produced during oxidative stress and enduring stimuli. As a matter of fact, native PAs are ROS scavengers
[73] and can play neuroprotectant roles, functioning as adaptive mechanisms maintaining homeostasis in CNS. Notably, the neuroprotectant effects of PAs focus on the roles of astrocyte–neuron crosstalk in maintaining healthy neuron–astrocyte network function. As already pointed out, while few adult neurons are synthesizing PAs but not holding PAs, astrocytes do not synthesize but collect PAs. Notably, the PA content declines during aging
[74], and the loss of PA homeostatic mechanisms may be relevant to AD, PD and other age-related diseases. Indeed, PA loss correlates with the development of CNS disorders, and PA restoration has a rescuing effect
[40]. The activation of the PA pathway and PA oxidation via SMOX or other oxidases, by generating ROS and by stimulating the antioxidant defence cell response (e.g., through the nuclear factor erythroid 2-related factor 2
[75]), may play both regulating and pathological roles, primarily in neurons and may shift the PA response from the protective, adaptive response-maintaining homeostasis towards a maladaptive detrimental mechanism. In fact, the increased PA levels following a short-term stimulus may have a beneficial role, while enduring stimuli such as repetitive brain trauma, cerebral arteriosclerosis-associated ischemia and metabolic stress lead to aberrant PA metabolism and, eventually, if it becomes maladaptive, to a deviant “PA stress response”, initiating the vicious cycle of neurodegeneration
[76]. The continuous induction of the PA pathway is followed by arginine brain deprivation, the extensive catabolic oxidation of PAs, ROS generation and the induction of oxidative stress
[76]. In neurodegenerative diseases, augmented PA metabolism results in the generation of H
2O
2 and reactive aldehydes including acrolein, which participate in the death of compromised tissue
[77]. Indeed, while SMOX activity in the healthy brain was found in some neurons, SMOX upregulation and overexpression was found in both PD
[78] and AD
[51] and in some CNS diseases
[79]. SMOX induction has been reported in diabetic retinopathy
[56] as well as in brain ischemia
[54]; in these pathological conditions, the upregulation of SMOX seems to be responsible for neuron damage (see
Figure 3).
Figure 3. Schematic representation of the major mechanisms involved in neuronal damage resulting from polyamine dyshomeostasis in the central nervous system. Polyamine dyshomeostasis-dependent mechanisms that have been suggested to play pivotal roles in representative relevant diseases are also highlighted. AD, Alzheimer’s disease; DR, diabetic retinopathy; HIV D, HIV-associated dementia; PD, Parkinson’s disease.
3.1. Alzheimer’s Disease
Starting from a pioneering observation of PA-altered levels in the autoptic brain of AD patients, where Spd was increased and Spm was reduced at the cortex level, it was suggested that abnormal PA activity may be involved in the neurodegenerative processes occurring in the brain of AD patients
[50]. Polyamines dysregulation was confirmed in AD brains, and altered levels of PA metabolic enzyme transcripts (including an increase in SMOX mRNA) were suggested to promote tau neuropathology and induce cognitive and affective impairments
[51]. PAs were proposed to trigger neurodegeneration in AD by condensing hyperphosphorylated tau
[80]. While the AD-associated PA response may be envisioned as an integrated part of the conserved adaptive mechanism, the prolonged induction of PAs possesses a limited efficacy in coping with gradual oxidative stress and may have detrimental effects due to toxicity issues. The continuous induction of the PA pathway with the extensive catabolic oxidation of PAs, ROS generation and the induction of oxidative stress may aggravate AD
[76]. Consistently, acrolein adducts are present in dystrophic neurites surrounding senile plaques
[81], acrolein levels are significantly increased in AD patients’ hippocampi
[82] and acrolein’s role in the AD pathogenesis has been suggested
[76][83]. In fact, AD offers an example of how PAs are “wonderful machinery” in the interplay between neuroprotective and detrimental effects, which can drive the development of disease. One of the key elements of AD is the accumulation of amyloid beta (Aβ). Amyloids are strongly charged anions, and PAs are cations, so PAs, which can function as scavengers of ROS
[73], may be bound by Aβ and neutralized or inactivated. Supplemental PA treatment is neuroprotective
[40][41][69][84][85][86], while the oxidation of PAs can cause neurodegeneration (
[87] and the current review). Therefore, glial cells, which are donors of PAs (unless Aβ or other acid proteins can buffer PAs), seem to be key players in the shift from synaptic function to dysfunction. In accordance, the dysregulation of astrocyte–neuron communication is considered to play major roles in neuron dysfunction in AD
[88]; reactive astrocytes have been suggested to be a “double-edged sword” in AD, exerting biphasic effects—beneficial or detrimental, depending on multiple factors
[89]. PAs may be one of these factors. As already mentioned, PA levels decrease with aging
[74]; the condition may be dramatically different in young and aging brains, as SMOX activity is different
[60], and recent evidence indicates that developing astrocytes can synthetize PAs, while adult astrocytes do not
[46].
3.2. Parkinson’s Disease
In PD patients, the Spm/Spd ratio in blood was significantly decreased, indicating the Spm conversion from Spd and a decrease in brain Spm; the Spm/Spd ratio enabled the discernment between PD patients and healthy controls
[90]. Interestingly, Spm was reported to prevent manganese-induced toxicity in dopaminergic neurons
[91], while SMOX, activating Spm breakdown and leading to the excessive formation of toxic aldehydes (such as acrolein), H
2O
2 and ROS were found to be up-regulated in PD
[92]. The role of PAs as gliotransmitters and regulators of neural function was suggested to be involved in triggering neuron oxidative stress and gliosis in PD
[92]. Due to the central role of PAs in cell functionality, it is not surprising that PA dysregulation destabilizes neuronal function. A growing body of evidence indicates that the cation-transporting ATPase 13A2 (ATP13A2) and PAs play a key role in the endo-lysosomal system and mitochondrial function, which are at the heart of neurodegenerative diseases. Several reports directly support the importance of PAs homeostasis, beyond ATP13A2, in neurodegeneration, particularly PD
[93]. ATP13A2 (also known as PARK9) is a lysosomal PA exporter with the highest affinity for Spm that promotes cellular PA uptake via endocytosis and PA transport into the cytosol
[93]. Notably, loss-of-function mutations in the
ATP13A2 gene seem causally linked to neurodegenerative diseases, including Kufor-Rakeb syndrome, a rare form of inherited juvenile-early onset Parkinson’s disease
[93][94][95][96][97][98][99][100][101], the while enhancement of ATP13A2 function has been proposed as a neuroprotective therapeutic strategy in Parkinson’s disease
[100][102][103][104]. ATP13A2-associated disorders are hallmarked by mitochondrial and lysosomal abnormalities; in fact, ATP13A2 might exert a dual protective effect by preventing the lysosomal accumulation of PAs and by increasing the PAs’ cytosolic levels. The contribution to the intracellular pool of PAs may mediate the protective effect of ATP13A2 on mitochondrial toxins and heavy metals, as PAs are well-established ROS scavengers
[73] and heavy metal scavengers and play essential roles in mitochondrial functionality and autophagy regulation
[44][93].
3.3. Diabetic Retinopathy and Retinal Pathologies
The neurotoxicity of the oxidative products of PA degradation such as H
2O
2, acrolein and aminopropanal led to the “aldehyde load” hypothesis for neurodegenerative diseases
[77][105]. Indeed, the inhibition of PA oxidases could prevent the NMDA-induced retinal neurodegeneration promoting cell survival, therefore providing a novel therapeutic target for retinal neurodegenerative disease conditions
[106]. As a matter of fact, the evidence points to roles for trauma, oxidative stress and PA metabolism alteration in optic nerve injury, glaucoma or prematurity retinopathy, also suggesting the regulation of PA metabolism as a neuroprotection strategy
[85][87][107]. Among PA oxidases, SMOX was reported to be involved in causing neurovascular damage in the retina
[56] and was proposed as a therapeutic target for neurodegeneration in diabetic retinopathy
[57]. Notably, the increased expression of SMOX was reported in the retina in response to hyperoxia-induced neuronal damage in retinopathy of prematurity
[87] and excitotoxicity-induced retinal neurodegeneration
[106]. The molecular mechanisms underlying neurodegeneration in the diabetic retina include the glutamate excitotoxicity and oxidative stress
[108]. Mitochondria are one of the major targets of oxidative insults, and oxidative damage-mediated mitochondrial dysfunction is a major mechanism for neuronal damage in neurodegenerative diseases
[109][110]. The proposed mechanism for SMOX-induced neuronal damage and dysfunction in diabetic retinopathy includes acrolein production, the depletion of antioxidants and mitochondrial dysfunction
[57].
3.4. HIV-Associated Dementia
The enzyme SMOX is also responsible for the chronic oxidative stress occurring in the brain tissues of human immunodeficiency virus (HIV)-infected patients, leading to the pathogenesis of HIV-associated dementia
[111]. In human neuroblastoma cell line HIV, the HIV-1 transactivator of transcription (Tat) elicits SMOX enzymatic activity upregulation through NMDA receptor triggering, thus increasing ROS generation, which in turn causes cell death
[111]. Furthermore, ROS produced by SMOX can stimulate the antioxidant defence cell response through the nuclear translocation of Nrf2 (nuclear factor erythroid 2-related factor 2), which induces the expression of the oxidative stress-responsive genes
[75]. These studies highlight that the NMDA/SMOX/Nrf2 pathway could be an important target for the protection against HIV-associated neurodegeneration
[75].