The Role of Glia in Stress: Polyamines and Brain Disorders

Synopsis

This review focuses on the roles of glia and polyamines (PAs) in brain function and dysfunction, highlighting how PAs are one of the principal differences between glia and neurons as they are surprisingly stored, but not synthesized, almost exclusively in glial cells from which they can be released to regulate neuronal synaptic activity. The review includes the novel role of PAs, such as putrescine (PUT), spermidine (SPD) and spermine (SPM) and their precursors and derivatives. However: (i) PAs have not yet been a focus of much glial research; (ii) PAs affect many neuronal and glial receptors, channels and transporters; (iii) PAs are therefore key elements in the development of many diseases and syndromes (iv) thus forming the rationale for PA and glia focused therapy for these conditions.

Glia versus neurons

Cajal¹ predicted how glia could help in health and disease by saying that glia are “insulating the neurons and switching their signaling”. His work has been analyzed by many scientists.^2–9 Cajal knew that glia were more than just connective tissue but could never prove this. He was able to highlight novel features of glial cells. These observations can be considered to be the discovery of the importance of glia as the “second brain.” Cajal who has been considered by many to be the “father of modern neuroscience” actually made a principal glial discovery: he visualized what are now known as radial glial cells (RGCs). Recent studies have shown that these cells are of ectodermic origin which means that RGCs are universal precursors for both neurons and glia. This broke the dogma that glia and neurons have separate origins and lineages.¹⁰ Then came the studies that neurogenesis was observed in adult human¹¹ and rat¹² brains shattering yet another dogma that neurogenesis was absent in the mature brain.

There is increasing evidence that RGCs build the brain by accommodating in the inner and outer subventricular zone (SVZ) in order to send their processes into the ventricular zone (VZ). These show polarity and are in fact the stem cells of the developing brain.^{13, 14,15,16}. Several morphologically distinct subtypes of RGCs in fetal macaque neocortex produce neurons and are guides for the migration of neural progenitors.¹⁷

Therefore, there are many different types of glial cells that are of RGC origin: NG-2, astrocytes, oligodendrocytes, tanycytes (in whole brain), Müller glia (in retina) and Bergmann glia (in cerebellum) as well as ependymal cells (in the ventricular surface). These cells represent the major neuroglial population in the adult central nervous system (CNS). On the other hand, peripheral glial cells such as Schwann cells, satellite glia (in the sympathetic, parasympathetic and sensory ganglia), enteric glia (in the ganglia of the digestive system), pituicytes (astrocytic glia in the posterior pituitary) are also types of neuroglia. Finally, while there are the microglia (mesodermal origin) that are macrophages in the brain, this review will not include the discussion of microglia.One of the major differences between glia and neurons is accumulation of biogenic polyamines; astrocytes expressing arginine decarboxylase can produce agmatine, a principal element in brain PA-exchange, and therefore, glial cells can be agmatine reservoirs.¹⁸ In general, the ratio of astrocytes to neurons (A/N) increases in evolution with increasing brain size¹⁹ and the highest glial cell (G) to neuron (G/N) ratio is found in brainstem²⁰ where the most important controls of body functions occur, for example control of respiration.^{21, 22} On the other hand, there is also evidence that the frontal cortex has the highest G/N ratio. RGCs, as well as astrocytes, are filled with the PAs spermine and spermidine.^{23, 24, 25} There is one surprising function of RGCs that was recently discovered which is that in the adult retina RGCs provide photon signaling and serve as light guiding fibers^{26, 27}.

Interaction of polyamines with receptors and ion channels

Polyamines such as SPD and SPM are involved in glial-neuronal communication, especially during periods of stress such as during ischemia and trauma; the mechanisms of storage and release are not well known. Since neurodegeneration is a major problem during stress, ischemia and CNS diseases, identifying potential neuroprotective mechanisms could provide new targets for therapeutic interventions. In the 1980s, it was discovered that the polyamine SPM was a principal radical group in spider venom^{28, 29, 30, 31} and the SPM portion of the venom could insert itself into and block glutamate receptors.³⁰ In the late 1990s, PAs were brought to the attention of neuroscientists. However the sources of PAs in the brain were not known.

PAs affect glial inwardly rectifying potassium (Kir)4.1 channels^{25, 32, 33} and most of the known neuronal receptors and channels.³⁴ In the brain and peripheral nervous system, SPM and SPD are known to have specific intra- and extracellular actions. SPM affects numerous receptors and channels in neurons with differing affinities ranging from ~10 nM – 200 μM.^{35, 36, 37, 38} Intracellular SPM/SPD induces voltage-dependent block of Kir channels^{25, 39, 40}, as well as neuronal nicotinic acetylcholine receptor channels^{35, 41}, glutamate (Glu)A-2 lacking α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) channels⁴², N-methyl-D-aspartate receptors (NMDARs)^{34, 43}, olfactory cyclic nucleotide-gated cation channels (CNGC)⁴⁴, and voltage-gated sodium channels.⁴⁵ In addition, some NMDAR and AMPAR channels show rectification in the presence of polyamines or their derivatives.^{31, 36, 42, 46, 47, 48, 49, 50} PAs are the strongest blockers of Kir channels, glutamate receptor-channels (such as AMPAR, NMDAR, KainateR) and acetylcholine receptor (AChR) channels^{31, 35, 36, 39, 51, 52, 53} and also act as the calcium-sensing receptors agonists⁵⁴ and antagonists of transient receptor potential cation channel, subfamily M, member 7 (TRPM7) and transient receptor potential cation channel, subfamily V member 1 (TRPV1) channels.^{38, 55} Also relevant are the extracellular actions of PA on GluA-2-lacking and GluA5/6-enriched AMPA/Kainate receptor channels in interneurons,^{56, 57, 58} because glial cells may release PAs to control synaptic activity. These receptor channels have an affinity for SPM in the micromolar (μM) range.

Spermine/Spermidine localization in CNS: Bi-directional polyamine signaling between glia and neurons

Surprisingly, the PAs spermidine and spermine are accumulated in glia (Fig. 1) and their distribution is clearly evolutionarily determined; it is found throughout the brain²³, retina^{24, 25}, peripheral nervous system⁵⁹ and in glial-neuronal co-cultures⁶⁰ of multiple species, including man²⁴. This phenomenon raises key questions: (i) What are the mechanisms that underlie such uneven distribution, accumulation and release from glia?; (ii) What are the consequences of PAs fluxes within the brain on neuronal function?; (iii) What are the roles of PAs in brain disorders and diseases?

Figure 1. Circulation of polyamines (PA) in brain.

A. Suggested interaction between astrocytes, neuronal dendrites and synapses and blood vessels based on bi-directional polyamine (PA) fluxes (i) between neurons and astrocytes, (ii) between astrocytes in their syncytium and (iii) between astrocytes and blood vessels. Polyamines are taken up and released from glia to neurons as well as propagated distantly through the syncitium (red arrows). B. Suggested PA pathways (uptake and release) in glia via connexin 43 (Cx43) hemichannels, Cx43 gap-junctions, reverse organic cation transporters (OCT) and vesicular release. C. Accumulation of spermine (SP) in astrocytes shown by immunocytochemical method in rat hippocampus. Astrocytes enwrap blood vessels and connect to each other. Note: no spermine and spermidine labels found in neurons in this stratum radiatum area of CA1 rat hippocampus.

Astrocytes enwrap pre- and post-synaptic neuronal terminals, generating a tripartite synapse.^{61, 62, 63} Failure of synaptic transmission^{64, 65} and vasodilation^{66, 67} has been ascribed to the malfunction of perisynaptic and perivascular astrocytes, respectively. Neuronal damage is evident after glial depletion in hepatic encephalopathy⁶⁸ and neuronal degeneration can occur following apoptosis of glial cells.⁶⁹ It has long been accepted that glia provide a support function to neurons by buffering extracellular K+ and glutamate.^{70, 71, 72, 73, 74} However, potential signal functions of glia are much less studied and understood.

Glial cells release signaling molecules such as arachidonate, glutamate, adenosine triphosphate (ATP), D-serine, tumor necrosis factor alpha (TNF-α) and others, which regulate neurons and blood vessels.^{61, 75–87} Thus, finding the signaling functions of glial cells and potential endogenous glial transmitters are one of the frontiers of glial research. Whilesuch gliotransmitters may be PAs theymostly underestimated and less studied; yet in the context of stress and glial function they might be a key element in helping to modulate neurons.

Polyamines may have harmful or neuroprotective effects via multiple pathways. For example, blocking calcium (Ca)²⁺-permeable GluR2-lacking AMPA receptor-channels by SPM ^{42, 56, 88, 89} reduces Ca²⁺ influx and prevents excitotoxicity.⁹⁰ Furthermore, potentiation by SPM of GluR-6 kainate receptors on inhibitory neurons^{58, 91} inhibits the activity of downstream pyramidal cells which can result in a neuroprotective effect. Ischemia, glucose deprivation or mild mechanical trauma all can result in neuronal death due to Ca²⁺ overload followed by apoptosis^{92, 93} if not protected by PAs.⁹⁰ The primary pathway for Ca²⁺ entry under these conditions is via fast GluR2-lacking AMPA receptors as well as slow NMDA receptors.^{50, 94, 95, 96} Therefore, the amount of PAs stored and released from glia may underlie the strength of neuroprotection. There might be an aging effect because the amount of PAs stored in the brain declines with age.^{97, 98, 99, 100, 101, 102} PA sensitive interneurons (lacking GluR2 subunits) within the brain, but not pyramidal neurons (expressing GluR2 subunits), are enveloped by PA-filled astrocytes (ex. in the hippocampus). During ischemic conditions there is a striking disparity in the rate of neuronal death in this region of the brain. Interneurons are spared, whereas pyramidal neurons are very susceptible.⁹⁰ This observation can be explained by the proximity of astrocytes as they take much better care of their direct neighbors, the interneurons. By contrast, the cortex interneurons and pyramidal cells are much less formally organized. Glial cells are unevenly distributed between the two cell types, making the neuronal sensitivity to ischemia less for interneurons than for pyramidal cells.¹⁰³. We suggest that the resistance of hippocampal interneurons to ischemia and apoptosis is a result of localized SPM/SPD release from the surrounding glia and the subsequent block of GluR2-lacking AMPA receptors. In support of this, there is compelling evidence for a protective role of exogenous PAs in brain ischemia and neurotransmitter-induced excitotoxicity: for instance the application of naphthylacetyl-spermine (NAS) or SPM/SPD in vitro or in vivo dramatically protects Cornu Ammonis area (CA)1 neurons under these conditions.^{90, 92, 93, 104} So, we suggest that when experiments are done with brain slices that might have their endogenous PAs washed out of the astrocytes by being disconnected from the blood circulation (a source of PAs) there should be supplementation with external PAs so as to keep PAs buffering the astrocytes which will improve neuronal survival and function.¹⁰⁵ This can be directly related to stress and neuroprotection. Another observed phenomenon is that PAs can be oxidized which results in the production of toxic compounds that can be captured by glial cells. Indeed, when cultured in the absence of glia, neurons show a delayed death in response to exogenous PAs.^{106, 107} Even though extracellular PAs are present in physiological tissues,¹⁰⁸ the healthy brain clearly avoids significant PA toxicity, probably because extracellular levels in the cerebrospinal fluid are buffered by the astrocytes. Consistent with this, and in contrast to the findings in pure neuronal cultures, 50 μM SPM applied exogenously in cortical brain slices is well tolerated, and neuronal death is not observed.⁵⁶ We, therefore, hypothesize that, in the intact brain, glial cells provide protection against trauma and excitotoxicity at least in part by using PAs as a buffer, thereby, preventing oxidation. PAs can then be released at appropriate sites with the potential to block inappropriate calcium permeability through glutamate receptor regulation. Indirect support for this hypothesis is the observation that in perfused brain slices, excitotoxicity induced by anoxia and toxic chemicals (ex. NMDA) in hippocampal slices was prevented if SPM was used.¹⁰⁹ If PAs may be considered as neuroprotective agents, then what are the physiological ranges of PA concentrations and what are the free versus bound forms of polyamines in the brain?

The total concentration of intracellular SPM in non-neuronal cells is high (3–10 milli [m]M)^{97, 110–114}, although SPM is greatly buffered in the cells by negatively charged phosphates, ATP, guanosine-5′-triphosphate (GTP), DNA, RNA, membrane proteins and other polypeptides.^{111,115, 116, 117, 118, 119} Since SPM and SPD are not synthesized in glia^{120, 121}, but instead are accumulated in glia^{23, 24, 25, 60} (Fig. 1C), the polyamines PUT, SPD and SPM may be exchanged during different conditions such as development, membrane depolarization, substrate co-transport or metabolic downregulation. Our immunolocalization data demonstrate preferential SPM/SPD accumulation in glia that enwrap blood vessels (Fig. 1). SPM and SPD were not found in the majority of neurons of the retina^{24, 25} and in brain.²³ This is consistent with our recent studies suggesting that the free SPM concentration in glia (up to 800 μM³³,) may be much higher than estimated in neurons ~10 nano(n)M – 80 μM.^{35, 36, 37, 38} In the brain, PAs show a strikingly uneven distribution. Using radioactive polyamines^{59, 116, 117} and polyclonal antibodies specific for PUT⁶⁰ or SPM/SPD^{23, 24, 25}, we and others have shown that PAs are taken up in brain^{110, 122,} and astrocytes are capable of taking up PAs.^{60, 123–128} (Fig. 1B).

Therefore, if PAs were to be synthesized in some neurons,^{129, 130, 131, 132} then they would be released from those neurons, possibly from synaptic neuronal vesicles, to the extracellular space. In fact, a vesicular transport system was found.^{133, 134} This leads to an immediate accumulation and storage of PAs in glial cells (Fig. 1A). An alternative source of PAs could be from the blood vessels, with which astrocytes maintain intimate contact, wrapping the vascular interface by endfoot processes (Fig. 1A, C). Here, glial cells may use several uptake pathways such as transporters and large pores (Fig. 1B) which will be discussed below.

Brain disorders and glia

Global amnesia, depression, stress, anxiety, autism, glioblastoma multiforme, glaucoma, migraines, neuropathic pain, sleeplessness and drug addiction are among a host of devastating neurological diseases or disorders for which prevention or a cure must be found.^135–152 We will show below that these disorders can be tightly linked with the PA machinery.

Since their original discovery by Leeuwenhoek¹⁵³, the polyamines SPM and SPD have attracted attention of scientists and clinicians.^{119, 154, 155} In the middle of the 20th century, there was great interest in their role in maintaining DNA structure and the possibility of treating cancer by blocking PA synthesis. Due, in part, to neurological complications in the anti-cancer treatment, this approach failed.¹⁵⁶ Later, studies have shown that this could have been predicted because of the existence of multiple effects of PAs on receptors and channels in the brain.

Indeed, multiple biological effects of PAs have been reported including increasing longevity,^{97, 101, 157, 158} cell proliferation and differentiation,¹⁵⁹ receptor and channel regulation,^{29, 34, 160, 161, 162} modulation of behavior, learning and memory,^163–167 as well as antinociceptive,^{38, 168, 169} neuroprotective,^170–172 antidepressant^{173, 174} and antioxidant effects^{97, 175}

While PAs are still a mystery in the brain^{18, 60, 110, 114, 122, 132, 157, 176–185}, they are known to be very tightly associated with glial cells. Altered PA metabolism may underlie certain brain disorders,^{186, 187} including depression with suicidal tendency.¹⁸⁸ Endogenous depletion of SPD and SPM by dietary means^{189, 190} or by genetic activation of SPD/SPM-acetyl transferase results in the loss of PAs and a loss of neuroprotection.^{191, 192} In spite of the fact that SPM/SPD are involved in the pathology of neurodegenerative diseases, they predominantly accumulate in glia and not in neurons.^{23, 24, 25} However SPD is not synthesized in glial cells^{120, 121, 179, 193} and thus, SPM cannot be synthesized without SPD, because it is the precursor. A solution to their origins comes from the idea that PAs are probably taken up from external sources by glial cells.^{114, 126, 194} It has been hypothesized that PAs are taken up by glia from the blood circulation, cerebral spinal fluid (CSF) or from macrophages penetrating the blood-brain barrier. One of the possible pathway is the organic cation transporter (OCT) system which is expressed in glia.^{195, 196}

Several CNS diseases have been shown to be associated with neuroglia such as astrocytes, oligodendrocytes, ependyma and other glial cells. Neurodegenerative diseases where glia play a key role are Alzheimer’s disease^{143, 197–200}, Amyotrophic Lateral Sclerosis ^201–203, Alexander disease²⁰⁴, Parkinson’s disease²⁰⁵, Huntington’s disease^{206, 207, 208}, multiple sclerosis ^{209, 210} and others.

Still other brain disorders and syndromes where glial cells play a pivotal role have been recognized. One of them is directly related to stress and epilepsy where (i) PA-dependent glial Kir4.1 channels are involved^{74, 211–214} together with (ii) glial connexin gap-junctions^{215, 216} and (iii) down-regulation of adenosine signaling.^{151, 217–219} In addition, there are severe disorders such as ischemia and stroke where reactive gliosis and an inability to regulate pH, K⁺-buffering, glutamate homeostasis and water exchange were found to result in the release of cytotoxic molecules, glial swelling and neuronal death.^220–224 The study of both physical^{221, 225} and chemical brain trauma resulting in the depression of glial metabolism⁶⁵ or reconstitution following brain edema and inflammation ^{226, 227} showed that reactive glia no longer were supporting neurons.

Intriguingly, the PA regulated Kir 4.1 channels^{228, 229} and a PA transporter OCT SLC22A subfamily, OCT3 are mislocalized²³⁰ in glial cells involved in brain cancer genesis (gliomas). These tumors produce increased intracranial pressure, metabolic deficiencies, toxicity and cell death.²³¹ Downs Syndrome and Snyder-Robinson Syndrome have malfunction of PA homeostasis.^{145, 232–234} The neurological manifestations of EAST/SeSAME syndrome with glial Kir4.1 mutations^{212, 214, 235} is solely of glial origin and blood-brain circulation disorders are seen when astrocytes cannot regulate vasodilation.^{67, 198, 236} All of these recent findings show the unique role of glial cells and PAs in CNS disorders.

By what mechanisms do glia accumulate and release PAs?

The enzymes ornithine decarboxylase (ODC) and spermidine synthase (SpdS) synthesize PT and SPD, respectively which are the precursors of SPM. In the normal brain, ODC and SpdS expression is typically found only in a few neurons, without any being detectable in glia.^{120, 121, 179, 237} In addition, we found that SPM-synthase is also absent in glia. This suggests under normal conditions, PAs may be synthesized outside glia in some neurons.^{120, 130, 131, 179, 238} Ultimately by unknown mechanisms, PAs will accumulate in glia.^{23–25, 60} Strong evidence for this is the finding that injections of radioactive putrescine into axons result in the transfer of the radioactive label to surrounding glial cells.⁵⁹ Even in brain areas where many adult neurons lack SpdS activity¹²⁰ and show low levels of SPM/SPD²³ there are still robust and potential sources of PAs in blood capillaries as PAs can permeate the walls of blood vessels¹⁸ to which glial cells are attached to by endfeet (Fig. 1 A, C). Therefore, accumulation of PAs in the glial cytoplasm is not primarily due to SPD/SPD synthesis, but instead to either passive or active transport (fluxes) of PAs (Fig. 1).

There are several potential pathways for exchange of SPM/SPD between glia, neurons and blood vessels such as the following:

1. large pores, including connexin (Cx) and/or pannexin (Panx) hemichannels,

2. ion channels

3. exocytotic vesicular release and endocytotic uptake (VRU).

All of these candidate pathways have been identified in glial cells for different molecules, but not yet for SPM/SPD.^{9, 78, 80, 239–244} It was suggested that PAs could be taken up from external sources by glial cells via hypothetical transporters which would bring PAs into the cells^{114, 126, 194,195} by transporters localized at the endfeet that are attached to blood vessels and brain ventricles. According to this view, the blood circulation and CSF would be the major sources of PA uptake. Finally, the macrophages that penetrate the blood-brain barrier and enter the brain may be additional sources.

Recent data show that SPD may be taken up by transporters such as polyspecific electrogenic organic cation transporters SLC22A subfamily¹⁹⁶ that were suggested as a pathway for monoamines such as dopamine, tetraethylammonium and others,^{126, 127, 194, 195} that are indeed also well suited for PAs.¹⁹⁶ Such transporters may function in a reverse mode releasing PAs by exchanging with other OCT substrates²⁴⁵, or, as it was shown for glutamate reverse transport²⁴⁶ or dopamine reverse transport^{247, 248} with similar and different mechanisms. Therefore, transporters may fulfill the function of regulating the PA content in the extracellular cleft and thus can regulate neuronal activity.

There may be other minor pathways for PAs; SPM permeates glutamate-receptor channels^{52, 53}, TRPV1 channels³⁸ and glial Kir4.1 channels.²⁴⁹ These are most likely negligible pathways for SPM flux since the channel pores (6 ångström (Å)-9Å) are comparable in size to SPM (4Å diameter and 16Å the length). Additionally, while exocytotic release as described for glutamate⁷⁸ and ATP²⁵⁰ may be applicable to PAs, there is no still evidence suggesting SPM release via this vesicular pathway.

The most likely major pathways for SPM/SPD exchange are via the large pores such as Cx and Panx hemichannels and transporters such as OCTs. Indeed, SPM/SPD may pass through connexin gap junctions and fill the astrocytic syncitium.¹⁰⁵ While several studies suggest that unpaired Cx43 hemichannels are present in the plasma membrane of astrocytes^{80, 240, 251} and in cell lines²⁵², Cx-hemichannels are normally closed due to the blocking effects of external divalent cations and voltage. When external calcium is decreased⁸⁰ or internal calcium is increased²⁵² Cx43-hemichannels open, forming large nonselective pores. Glial cells express a number of connexins, the most prevalent is Cx43, which together with Cx26, Cx30 and Cx45, form glial hemichannels and gap junctions with a large pore diameter (10Å-15Å) ^{253, 254} which SPM can readily pass through. Since Cx38 hemichannels in Xenopus oocytes are permeable to SPD²⁵⁵ and that there are hemichannels in rat cortical astrocytes, most probably Cx43, permeable to PAs¹⁰⁵, then this makes the Cx-pathway a likely candidate to be where uptake and release of SPM are regulated. Consistent with this possibility is the finding that gap junctions comprised of Cx43 are not blocked by PAs, while others made from neuronal Cx40 are blocked by SPM.^{256, 257} An alternative potential pathway for SPM/SPD fluxes are Panx-hemichannels, three of which (Panx1-3) are expressed in glia. These are nonselective pores that can open with normal external calcium levels.^{243, 251} However, Panx is not blocked by gadolinium²⁵⁸, while SPM flux in astrocytes is blocked by gadolinium making Panxs an unlikely pathway.²⁵⁹ In support of a direct exchange of SPM between the neuronal and glial cytoplasm via Cx hemichannels is the finding that much larger dye molecules can pass between astrocytes, but not between oligodendrocytes and neurons or between neurons.²⁶⁰ We have found that SPM/SPD fluxes are Ca²⁺- and -gadolinium (Gd³⁺) sensitive, favoring the possibility that Cx hemichannels rather than Panx hemichannels are the relevant pathway.^259–262

As mentioned above, another potential mechanism for SPM/SPD fluxes in glial cells is through polyspecific cation transporters (OCTs).¹⁹⁶ These transporters can translocate monovalent, divalent and even polyvalent cations such as SPM and SPD.^{196, 263} Polyspecific cation transporters (which include OCT1, OCT2, OCT3 and OCTN2, multidrug and toxin extrusion protein or MATE) are present in astrocytes and control signal transmission and energy homeostasis by removing released transmitters and substrates, such as dopamine, norepinephrine, epinephrine, 5-hydroxytryptamine, carnitine and histamine, from the extracellular space.^264–267 OCT transporter mRNA has been found in cultured astrocytes^{195, 268, 269} and is well described.^{196, 270}

Glia-controlled polyamine regulation in the neuronal network

There should be a mechanistic basis to explain diseases, disorders and syndromes that are associated with glia and with PAs. The glial membrane potential (~ −85 mV) is typically 20–30 mV more hyperpolarized than resting neurons, therefore, polyvalent cations, SPM⁴⁺ and SPD³⁺ will be concentrated in glia by electrodiffusion via large pores (like Cx-43) or OCTs (which require electrical transmembrane potential) due to this hyperpolarization. Once inside the cell, PAs will be buffered by polyanions (RNA, ATP, acid proteins, etc.) in the cytosol. When neurons are generating action potentials, there can be large transient falls in extracellular calcium ([Ca²⁺]_o)^{271, 272} and sodium ([Na⁺]_o), and a rise of potassium ([K⁺]_o)²⁷³, all of which will accelerate both glial depolarization and hemichannel opening. Intriguingly, with a delay of only a few minutes during ischemia, there is a dramatic increase of [K⁺]_o (to 55 mM) and a lowering of [Na⁺]_o, [Cl⁺]_o, [H⁺]_o and [Ca²⁺]_o- to 60 mM, 75 mM, pH=6.5 and 0.08 mM from normal levels, respectively.^274–277 This provides favorable conditions for Cx43-hemichannel opening that will not block OCTs. Normally, [Ca²⁺]_o can block Cx43 hemichannels.^{80, 252} During epilepsy, ischemia, spreading depression or trauma, the activity of [Ca²⁺]_o is decreased by an order of magnitude from 1.2 to 0.06 mM.^274–277 This is a condition under which Cx43 hemichannels can open,^{278, 279} potentially allowing PAs to be released outside glia where they may help to remove the external H⁺ block of Cx43. Also, during ischemia, pH may reach acid levels of ~6.5–6.1^{274, 276} and OCTs are depressed by acid pH and depolarization.^{263, 270, 280} Therefore, the Cx43 and OCT pathways function differently: at normal conditions OCTs are a major PA pathway, then with excessive neuronal activation or ischemia there is a stimulation that opens Cx43 hemichannels that will allow the release of PAs, thus, conferring neuroprotection as mentioned above by blocking Ca²⁺-permeable neuronal channels and preventing apoptosis.

Selective depolarization of astrocytes with aminoadipic acid changes the neuronal firing rate; this is blocked by carbenoxolone, a non-selective blocker of hemichannels.²⁶⁰ This is consistent with the idea that hemichannels are key players in neuronal regulation. SPM/SPD release via hemichannels from astrocytes may be induced by depolarization (Fig. 1B). The consequences of such release will depend critically on which neuronal receptors are exposed to the potentially high localized release.

Polyamines: role in the CNS disorders

While previously unrecognized, recent data highlight dynamic signaling within glia and between glia and neurons via PAs. Many studies have reported neuroprotective effects of PAs.^{92, 109, 121, 175, 187, 281, 282} Under pathological conditions, PAs may be oxidized and converted to cytotoxic aldehydes and reactive oxygen species, which may be responsible for subsequent neurotoxic damage.^{197, 281} The down-regulation of the synthesis of SPM due to a mutation in the X-chromosome, causes Snyder-Robinson Syndrome that so severely affects human brain function that the result is mental retardation, hypotonia and cerebellar dysfunction.^{233, 234} Conversely, unmodified PAs may block Ca²⁺-permeable receptors and channels which is critical in protecting neurons from apoptosis (see above). Despite these recognized modifications and potential targets, the exact localization and dynamics of PAs in the brain are largely unknown.

What is important to know is that many of the aforementioned disorders show a tight link between PAs and glia. SPM/SPD levels decline with age^{97, 101, 158} and are involved in Parkinson’s disease.^{283, 284} These reports clearly demonstrate a vital role for PAs in brain plasticity. For instance, there is evidence that in brain diseases PAs play a principle role in restoring age-related memory impairment¹⁰¹ as there is evidence that PA depletion stops mammalian cell growth and PA supplementation will reverse the effect.²⁸⁵ Treatments with SPD, agmatine, and by genetic modulation that increase endogenous PA levels resulted in not only an increase in life span⁹⁷, but also in memory restoration.^{97, 101, 199} PA-rich nutrition increased the life span of aging mice^{98, 99, 100, 286} while some amines and PAs (such as agmatine, arginine, putrescine, spermidine, spermine) are lost during aging and in the diseased CNS.^{97, 102, 157, 179–182, 286} Also, PAs inhibit age-associated changes in global DNA methylation as well as dimethylhydrazine-induced tumor genesis.¹⁰⁰ It has been hypothesized that information in the brain can be stored (memory) in the chromatin (a complex of DNA-proteins-polyamines) and can be changed during epigenetic chromatin modifications as in Alzheimer’s disease where macrophages penetrate the brain via the “leaky” blood-brain-barrier and probably cleave up “memory-proteins” in chromatin.²⁸⁷ Since, glia (G) outnumber neurons (N) in the human brain (the G/N ratio reaches 11.35 in brainstem and 3.76 in cortex²⁰)) glia can represent a substantial source of memory capacity especially given that PAs are bound to DNA in chromatin, to RNA and to acid proteins.^{111, 167, 288} There are severe psychiatric disorders directly linked with neuroglia and PA dysfunction such as schizophrenia and mood disorders, where failure of PA exchange is seen^179–182, autism with disorders in the PA and Na-H-exchange,^{149, 289} depression which is associated with decreased glial cell mass, OCTs and PA levels,^{173, 290} suicide which correlates with PA homeostatic imbalances,²⁹¹ aging where glial cells lose PAs^{99, 100, 286} and downregulate Ca²⁺-signaling²³⁶ and fear extinction where PAs reinforce extinction via NMDAR regulation.²⁹² Recently, many human CNS diseases have been linked with PA levels including Alzheimer’s disease^{102, 293, 294}, Parkinson’s disease^{128, 295, 296, 297}, Huntington’s disease^{167, 298–301} and Amyotrophic lateral sclerosis.^{186, 187, 302} In spite of a broad phenomenological description in the literature, there still remain functional links between PAs and glia in the brain that have not been mechanistically deciphered. However, what is known is that PAs are ultimately stored in glial cells with the level of storage and PA buffering capacity depending solely on glial cells.

Therefore, research is focusing on glial-based PA-related CNS problems for several reasons. First, PAs are very powerful modulators of the neuronal-glial network.^{56, 303} Second, PAs are accumulated preferentially in glial cells not in neurons.^{23, 24, 25} Third, PA homeostasis is critical for life.^{97, 101,184, 185, 304} Fourth, PA-exchanges play a principal role in brain disorders including inflammation^{224, 226, 227}, depression²⁹⁰, anxiety^{305, 306} and others (see above). Fifth, PA-sensitive glial Kir4.1, Cx43/Cx30 and OCTs are fundamental for survival of the cells and whole brain.^{29, 74, 126, 127, 209, 213, 215, 223, 230, 249, 259, 270, 307} Sixth, Cx43 and OCTs are candidates for PA uptake and release.^{105, 196, 259, 261, 245, 262} Finally, because CNS disorders are closely related to PA exchange in stress and aging^{102, 170, 157, 179–182} therapeutic supplement by PAs or their precursors have been suggested for neuroprotective treatment.^{18, 98–102, 110, 286, 308, 309} As for the supplemental therapy, attention should also be focused to PA-carriers such as Cx43 and OCT proteins. While there has been an increase in the research on PAs, the mechanisms of glial dependent PA actions in the CNS is just beginning to be elucidated and it seems that the glial-PA avenue for research has not yet been well blueprinted as more work needs to be done.

Conclusion

As is now evident, PAs are stored in glia^23–25 and together with their derivatives and precursors have been suggested as neuroprotective agents.^{18, 60, 93, 109, 110, 122, 175, 177, 194, 282, 308} However, the oxidation of PAs results in toxic products such as aldehydes that may be harmful for brain cells.¹⁰⁷. Lower molecular weight polyamines have less toxicity. One of the PA precursors, agmatine (decarboxylated arginine or 1-(4-aminobutylguanidine) is a natural product discovered over a 100 years ago in herring sperm³¹⁰ and research indicates its exceptional modulatory action at multiple molecular targets, including neurotransmitter systems, nitric oxide and PA production, therefore providing bases for current therapeutic applications. A triamine (SPD) and a diamine (PUT; 1,4-diaminobutane) were isolated from pro- and eu-karyotic systems; thereby showing that there are many sources of different polyamines in any diet.^{98–101, 311, 312} Recent preclinical and initial clinical evidence show a very positive effect of agmatine treatment¹⁸ making for challenging research opportunities for the use of agmatine in treating diabetes, neurological trauma, neurodegeneration, different types of addictions, mood disorders, cancer, cognitive and memory disorders. It seems that agmatine (amino-guanidine), guanidine based drugs and synthetic amines are examples of the best substrates for organic cation transporters of the SLC22A subfamily^{196, 313, 314} which have been found and characterized in astrocytes and other glial cells in the brain.^{263–270, 280} Specifically, recent data show that a polymorphism of OCT1 is directly related to Parkinson’s disease³¹⁵ and OCT3 is a key modulator of neurodegeneration in the nigrostriatal dopaminergic pathway.³¹⁶

Therefore, not only should agmatine, SPD and other polyamines be a focus in designing future pharmaceutical tools to treat psychiatric disorders but also the PA transporter systems (for example OCTs^{315, 316}, Cx43¹⁰⁵ ) and buffering agents for PAs such as acid proteins, ATP, phosphates, etc.¹¹¹ should be considered.

Finally, targeting of certain disorders with PAs needs to be taken with great care as for example in the case of gliomas where translocation of glial Kir 4.1 and OCT3 from the glioma cell plasma membrane to the nuclear membrane^228–230 makes these very important glial proteins unreachable by drugs from the extracellular space. The future challenge will be to find drugs that are able to reach these crucial areas and modulate the PA system. Only this way will we be able to know if the modulation of the PA system can give us another tool in the fight against brain dysfunction.

Key Points.

1. Polyamines (Pas) are one of the principal differences between glia and neurons as they are surprisingly stored, but not synthesized, almost exclusively in glial cells from which they can be released to regulate neuronal synaptic activity.

2. PAs have not yet been a focus of much glial research.

3. PAs affect many neuronal and glial receptors, channels and transporters.

4. PAs are key elements in the development of many diseases and syndromes, thus forming the rationale for PA and glia focused therapy for these conditions.

Glossary

μM

micromolar

Å

ångström

A/N

ratio of astrocytes to neurons

AChR

acetylcholine receptor

AMPAR

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor

ATP

adenosine triphosphate

Ca

calcium

CA

Cornu Ammonis area

Ca²⁺

extracellular calcium

Cl⁺

extracellular chloride

CNGC

cyclic nucleotide-gated cation channels

CNS

central nervous system

CSF

cerebral spinal fluid

Cx

connexin

G G/N

ratio of glial cells to neurons

Glu

glutamate

GTP

guanosine-5′-triphosphate

H⁺

extracellular hydrogen

Ir K

Kir, inwardly rectifying potassium

K⁺

extracellular potassium

m

milli

MATE

multidrug and toxin extrusion protein

n

nano

Na⁺

extracellular sodium

NAS

naphthylacetyl-spermine

NMDAR

N-methyl-D-aspartate receptor

OCT

organic cation transporter

ODC

ornithine decarboxylase

Panx

pannexin

PAs

polyamines

PUT

putrescine

RGCs

radial glial cells

SPD

spermidine

SpdS

spermidine synthase

SPM

spermine

SVZ

subventricular zone

TNF-α

tumor necrosis factor alpha

TRPM7

transient receptor potential cation channel, subfamily M, member 7

TRPV1

transient receptor potential cation channel, subfamily V, member 1

VRU

vesicular release and endocytotic uptake

VZ

ventricular zone