Acid-Sensing Ion Channels: Focus on Physiological and Some Pathological Roles in the Brain

Abstract

Acid-sensing ion channels (ASICs) are Na⁺-permeable ion channels activated by protons and predominantly expressed in the nervous system. ASICs act as pH sensors leading to neuronal excitation. At least eight different ASIC subunits (including ASIC1a, ASIC1b, ASIC2a, ASIC2b, ASIC3, ASIC4, ASIC5) are encoded by five genes (ASIC1-ASIC5). Functional ASICs assembled in the plasma membrane are homo- or heteromeric trimers. ASIC1a-containing trimers are of particular interest as, in addition to sodium ions, they also conduct calcium ions and thus can trigger or regulate multiple cellular processes. ASICs are widely but differentially expressed in the central and peripheral nervous systems. In the mammalian brain, a majority of neurons express at least one ASIC subunit. Several recent reviews have summarized findings of the role of ASICs in the peripheral nervous system, particularly in nociception and proprioception, and the structure-function relationship of ASICs. However, there is little coverage on recent findings regarding the role of ASICs in the brain. Here we review and discuss evidence regarding the roles of ASICs: (i) as postsynaptic receptors activated by protons co-released with glutamate at glutamatergic synapses; (ii) as modulators of synaptic transmission at glutamatergic synapses and GABAergic synapses; (iii) in synaptic plasticity, memory and learning; (iv) in some pathologies such as epilepsy, mood disorders and Alzheimer's disease.

1. INTRODUCTION

Acid-sensing ion channels (ASICs) are proton-gated voltage-independent ion channels. ASICs belong to the epithelial sodium channel/degenerin (ENaC/DEG) family of amiloride-sensitive ion channels [1]. The history of ASICs began in 1980 when currents activated by extracellular acidification in the neuronal membrane were first reported [2], and the existence of a specific proton-activated receptor was postulated. The concept of a highly selective Na⁺ current gated by specific receptors for protons was not readily accepted at that time [3]. In 1997 the proton-activated receptor was cloned and named the “acid-sensing ion channel”. This heralded a new era in the investigation of the ASICs [3]. It has been established later that there were at least eight different ASIC subunits (including ASIC1a, ASIC1b, ASIC1b2 ASIC2a, ASIC2b, ASIC3, ASIC4, and ASIC5) encoded by five different genes [4-8]. Unlike ENaCs, ASICs form homo- or heterotrimeric protein complexes representing channels with distinct properties [9-11].

Well-known potassium-sparing diuretic amiloride was the first described molecule blocking ASICs [12]. However, amiloride is a non-selective blocker because it also affects ENaC [13] cyclic GMP-gated cation channel [14] sodium-hydrogen antiporter 1 [15], T-type calcium channel [16] and urokinase plasminogen activator [17]. For many years, no new ASIC ligands have been reported until the discovery of psalmotoxin, a potent and highly selective natural antagonist of ASIC1a homomeric channels [18]. This invention initiated extensive research of the pharmacological properties of ASIC channels and their role(s) in the brain function. Several pro-inflammatory endogenous neurotransmitters have been found to affect brain ASIC functioning and enhance their response to acidification (arachidonic acid and RF-amide peptides [19], spermine [20], dynorphins [21], histamine [22-24]. These findings are very significant for understanding the mechanisms of neuronal damage caused by ASICs since, without these factors, extracellular acidification subsequently leads to the cessation of ion flux into the cell due to the desensitization of ASIC channels. Twelve years after the invention of psalmotoxin, the same group isolated new toxins from black mamba venom, called mambalgins, able to block both homomeric ASIC1a and its heteromers [25]. Recently, based on the binding of psalmotoxin to the acidic pocket of the ASIC, we have found a potent small-molecule inhibitor of ASIC1a channels, but, unlike psalmotoxin, it demonstrates a strong dependence of the blocking action on pH [26]. To the best of our knowledge, it is the most potent small-molecule inhibitor of ASIC1a to date.

ASICs are widely expressed in both peripheral and central nervous systems [27, 28], having numerous roles in a variety of important physiological functions and pathological states. There are at least eight Na⁺ channels encoded by five different genes. The proteins encoded by these genes represent subunits that form homo- and heterotrimers with different biophysical and pharmacological properties [9, 29-31]. Splice variants of the ASIC1 gene encode ASIC1a, ASIC1b and ASIC1b2 subunits; ASIC2 – ASIC2a and ASIC2b subunits [32, 33]. Splice variants for ASIC3, ASIC4, and ASIC5 have not yet been identified. Within the nervous system, ASIC1a, ASIC2a, and ASIC2b [6, 7, 34] are primarily expressed in the CNS, while ASIC3, ASIC1b, and ASIC1b2 are in sensory neurons of the peripheral nervous system [35, 36]. The expression of ASIC3 correlates with its role in multimodal sensory perception, including nociception, mechanosensation, and chemosensation. Although ASIC3 is present in some specialized areas of the brain [37, 38], ASIC1a, 2a, and 2b are the major subunits, as determined by the expression [6, 7, 27, 34], immunohistochemical [39, 40] and electrophysiological studies [41, 42]. These subunits form functional (i.e. can be activated by physiological changes in pH) homo- and heterotrimeric complexes [28]. ASIC1a homomers are especially sensitive to pH (Fig. 1). The other ASIC subunits (and heterotrimers) are less sensitive to protons, and ASIC4 and ASIC5 are not activated by protons at all [4, 43-45]. It is more intriguing that, unlike ASIC2a, heterologous expression of ASIC2b alone does not support proton-gated currents [46]. Of particular interest is an overlap in the expression of ASIC2 and ASIC1a proteins, with synaptically dense regions in several brain structures [40] highlighting an important role of ASICs in synaptic function.

Fig. (1).

Gating properties of ASIC1a channel. ASIC1a channel is the principal sensor of protons in the brain. At pH values greater than 8.0, virtually all of these channels are closed and impermeable to ions. Rapid acidification to pH 5.0 results in the appearance of transient ion current mediated mainly by sodium ions (middle panel). However, some data suggest that homomeric ASIC1a channels could also conduct calcium-inducing strong deteriorating effects on neurons. This current activates very fast (<1 ms) and, in turn, falls to a zero level in 2-4 seconds. Unlike ASIC3, ASIC1a does not conduct a steady-state current due to the transition from an open into a non-conducting desensitized state in the continuous presence of excessive protons. Upon acidification removal, the ASIC1a channel slowly recovers into closed, sensitive to protons state (left panel). ASIC1a channel starts its activity at pH 6.9 while decreasing the pH to values less than 6.0 does not provide any additional current (right panel). When the background pH falls to values less than 8.0, the ASIC1a channels begin the transition from a closed to a desensitized state, and at pH 7.0, virtually all of the channels are desensitized. The essential part of brain ASIC1a channels is being desensitized even at normal physiological pH of 7.4. However, several endogenous substances decrease the desensitization of ASIC1a, allowing them to be active even at relatively low pH, which is very relevant for some pathological states such as ischemia, etc. (A higher resolution/colour version of this figure is available in the electronic copy of the article).

Virtually all known ASIC subunits are present in the mammalian brain (Table 1), except the ASIC1b [39] and ASIC1b2 [47] subunits which appear to be specific for the peripheral nervous system. ASICs are predominantly expressed in the brain neurons, however, several studies have documented the presence of ASICs in glial cells. In particular, the expression of ASICs has been shown in oligodendrocyte-precursor cells (also known as NG2 cells) [48, 49]. ASIC1a, ASIC2a, and ASIC4 mRNAs were found in oligodendrocyte lineage cells, ASIC transcripts were also detected in brain white matter, and ASIC1a protein expression was detected in white matter oligodendroglia [48]. The existence of ASIC1, ASIC2a, and ASIC3 has been demonstrated in microglia using quantitative real-time PCR (qPCR), western blotting, and immunofluorescence experiments [50]. Expressions of microglial ASIC1 and ASIC2a were upregulated by lipopolysaccharide stimulation (that enhances the production of inflammatory mediators) and this could be inhibited by blockers of ASICS [50]. The presence of ASIC1, ASIC2a, ASIC3 has also been reported in astrocytes [51]. Under physiological conditions, most of the ASICs in astrocytes were expressed in the nucleus [51, 52]. On the other hand, in reactive astrocytes (in the hippocampus of epileptic mice), Ca²⁺ influx, induced by acidification, was sensitive to PcTX1 [52], indicating the presence of functional ASICs. Glioblastoma cell lines express functional acid-sensing ion channels ASIC1a and ASIC3 [53]. Thus, it appears that substantial expression of ASICs in glial cells is mostly related to

Table 1.

A summary of different ASIC subtypes cloned to date with their localization in the mammalian brain.

Gene	Subtypes	Other Names	Localization
ASIC1a(Accn2)	ASIC1a	ASIC, ASICα, BNaC2α	Amygdala [57, 58, 40] 59], hippocampus [42, 40, 60, 61, 62], cerebellum [40, 60, 63], olfactory bulb [64, 40], striatum [65-68, 40], cortex [65, 57], [41, 69, 70, 40], [60, 63], brain stem [40, 60], brain [41, 39, 62, 71, 72]
	ASIC1b	ASICβ, BNaC2β	N/A
	ASIC1b2c	ASICβ2	N/A
ASIC2a(Accn1)	ASIC2a	ASIC2, MDEG1, BNaC1α, BNC1α	Amygdala [40], hippocampus [40, 42, 60, 61] cerebellum [40, 60], olfactory bulb [40], striatum [67, 68], cortex [69, 40, 60, 73] brain stem [40, 60], brain [40, 61, 62, 71]
	ASIC2bc	MDEG2, BNaC1β
ASIC3a(Acccn3)	ASIC3b	DRASIC, TNaC1	Brain stem [60, 74], cortex [60], cerebellum [60], brain [71] hippocampus [60]
ASIC4a(Accn4)	ASIC4b,c	SPASIC, BNaC4	Amygdala [75], cortex [60], brain stem [60], brain [43, 4, 75, 41], striatum [75], pituitary gland [4], cerebellum [75], thalamus [75, 76], hypothalamus [76], olfactory bulb [75], hippocampus [75]
ASIC5a(Accn5)	ASIC5b,c	BLINaC, INaC, BASIC	[8]

^aGene name recommended by the HUGO Gene Nomenclature Committee.

^bAt least 2 spliced variants have been reported from mouse Accn3, Accn4, and Accn5 genes in various mRNA databases [32], but not cloned yet.

^cInsensitive to protons

pathology. ASICs are also expressed in vascular smooth muscle [54] and immune cells [55, 56]; nevertheless the level of ASICs expression in these cells is typically lower.

Several recent reviews have summarized the role of ASICs in the peripheral nervous system, in particular, in nociception and proprioception, [77, 78], as well as the structure-function relationship of ASICs [11, 79] and their evolution [80]. However, there is little coverage on recent findings regarding the role of ASICs in the brain.

After providing a brief background, this review will focus on: (i) the physiological role of ASICs in the brain as the basis of the better understanding of their role in (ii) different brain disorders. We will review the evidence regarding the role of ASICs: (i) as postsynaptic receptors activated by protons co-released with glutamate at glutamatergic synapses; (ii) as modulators of synaptic transmission at glutamatergic and GABAergic synapses; (iii) in synaptic plasticity, memory and learning; and (iv) in pathologies such as epilepsy, mood disorders and Alzheimer's disease.

2. PHYSIOLOGICAL ROLES OF ASICS IN THE BRAIN

2.1. Roles of ASICs in Synapses

Under physiological conditions, the brain’s extracellular pH is reasonably constant. However, several mechanisms can induce transient and localized pH fluctuations [23, 81]. ASIC1a, 2a and 2b are the major subunits expressed in the brain [28]. ASIC1a is of particular importance as it can be activated by acidosis in the physiological range. In contrast to other ASICs, which are permeable only to sodium ions, ASIC1a also conducts calcium ions; thus its activation can trigger/regulate multiple cellular processes.

ASICs, especially those expressed in the brain, are typically activated by a rapid drop of pH, and they rapidly become desensitized. Slow acidification results in steady-state desensitization without inducing detectable currents [82, 83], (Fig. 2). Thus, there are actually very few areas in the brain where ASICs can be activated under physiological conditions. The synaptic cleft is one of these areas [84-87] (but see [88] for conflicting evidence). Rapid acidification occurring during synaptic vesicle release can activate ASICs on both pre- and postsynaptic neurons (Fig. 3). In the latter case, a fraction of postsynaptic current is mediated by ASICs, with protons acting as a co-transmitter. Additionally, activation of ASICs can modulate synaptic strength by affecting transmitter release and/or sensitivity of postsynaptic receptors.

Fig. (2).

Slow acidification results in a steady-state desensitization of ASIC-currents. In the experiments illustrated [83] ASIC-currents were evoked by acidifications with a similar amplitude (from pH 7.4 to pH 6.0) but with different rise times (see [83]) for more details). Upper panel. pH of the applied solution measured using pH-sensitive fluorescent dye; 10–90% fall-time of the fluorescence change corresponds to the interval between the two squares (corresponding values are shown above the traces). Lower panel. ASIC-currents evoked by these acidifications. Notice the striking reduction of the amplitudes of ASIC-currents evoked by slower acidifications. Adapted from [83] (part of Figure 4) published under the terms of the Creative Commons Attribution License (CC BY). Copyright^© 2020 Alijevic, Bignucolo, Hichri, Peng, Kucera and Kellenberger. (A higher resolution/colour version of this figure is available in the electronic copy of the article).

Fig. (3).

Potential roles of ASICs in synaptic transmission and its modulation. Rapid acidification occurring during synaptic vesicle release (1) can activate ASICs on both pre- and postsynaptic neurons (2). In the latter case, a fraction of postsynaptic current is mediated by ASICs, with protons acting as a co-transmitter. Additionally, activation of ASICs can modulate synaptic strength by affecting transmitter release (3) and/or sensitivity of postsynaptic receptors (4). The scheme represents a generalized synapse. NT – neurotransmitter (for instance, glutamate or GABA); “main“ receptor - postsynaptic receptor (AMPA and NMDA receptors in glutamatergic synapse, GABA_A receptor in GABAergic one). (A higher resolution/colour version of this figure is available in the electronic copy of the article).

Usual strategies to investigate potential roles of ASICs in synaptic transmission are a) examining the effects of ASICs ‘silencing’ on synaptic transmission using genetic and pharmacological tools; b) studying whether these effects are due to endogenously occurring acidification. In regard to the latter, the effects are compared in extracellular solutions with different concentrations of pH buffer [58, 66, 89, 90].

2.1.1. ASICs as Postsynaptic Receptors Activated by Protons at Glutamatergic Synapses

2.1.1.1. Synapses on Pyramidal Neurons in Lateral Amygdala

Convincing evidence indicating that postsynaptic ASICs are activated during synaptic transmission and mediate a fraction of the excitatory postsynaptic currents (EPSCs) at glutamatergic synapses on pyramidal neurons in lateral amygdale was reported in 2014 [58]. Indeed, in wild-type animals, block of ionotropic glutamate receptors revealed ‘residual’ EPSCs which were subsequently blocked by antagonists of ASICs. No ‘residual currents’ were observed in neurons from ASIC1a^−/− animals. Additionally, it has been shown that: (i) presynaptic stimulation evokes local proton transients; and (ii) changing pH buffer capacity of the extracellular solution alters ASIC-dependent EPSCs. These results further

support the idea that protons act as a neurotransmitter at glutamatergic synapses in lateral amygdala. Importantly, despite the small (~5%) relative contribution of the ASIC-dependent component to postsynaptic currents in amygdala neurons, ASIC1a is crucial for long-term potentiation (LTP) induced by high-frequency stimulation (HFS). Indeed, similar to the ASIC-dependent component of EPSCs, LTP was modulated by changing the capacity of extracellular pH buffer and was absent in ASIC1a^−/− animals.

2.1.1.2. Synapses on Ventral Striatum (Nucleus Accumbens) and Dorsal Striatum Neurons

It has been shown that residual evoked EPSCs, recorded in the presence of blockers of ionotropic glutamate receptors (AP5 and CNQX), were nearly eliminated in ASIC1a^−/− mice. These residual currents were restored by reinstating ASIC1a expression in the nucleus accumbens with adeno-associated virus that contained the coding sequence of ASIC1a [66]. As expected for responses mediated by endogenous acidification, ASIC-dependent postsynaptic currents were increased when neurons were bathed in extracellular solution with reduced pH buffering capacity, while increasing the buffering capacity reduced the ASIC-dependent current [66]. Additionally, inhibiting or deleting carbonic anhydrase IV, an enzyme critical for regulating extracellular pH buffering in the brain, increased ASIC-dependent EPSCs in the nucleus accumbens [66]. Besides being directly responsible of a fraction of postsynaptic currents in neurons of nucleus accumbens, postsynaptic ASICs also modulate synaptic responses mediated by other postsynaptic receptors. Indeed, the loss of ASIC1a (i) increased inward rectification of AMPA-mediated EPSCs and (ii) changed the AMPA-to-NMDA ratio of EPSCs [66]. Both effects were reversed by re-establishing ASIC1a expression. Interestingly, ASIC1a disruption increased dendritic spine density in nucleus accumbens [66], while the opposite effect was observed in the hippocampus [91].

It has been demonstrated more recently that postsynaptic ASICs also mediate a fraction of EPSCs at glutamatergic synapses on medium spiny neurons (MSNs) in the dorsal striatum [92]. Notably, ASIC1a has a similar role in synaptic remodeling in dorsal and ventral striata. Namely, in Asic1a KO MSNs, increased dendritic spine density and decreased NMDAR/AMPAR ratio were exhibited in both regions [92]. In the dorsal striatum, however, ASIC1a-dependant synaptic remodeling regulates procedural motor learning [92], while in the nucleus accumbens, it affects cocaine-seeking behavior [66].

2.1.1.3. Synapses on the Medial Nucleus of the Trapezoid Body (MNTB) of the Auditory System (Calyx of Held–MNTB Synapse)

A fraction of EPSCs revealed in the presence of blockers of AMPA, NMDA, GABA and glycine receptors, is blocked by antagonists of ASICs (amiloride and PcTx-1) and is absent in knockout mice for the ASIC-1a subunit (ASIC1a^-/- [89]. High-frequency stimulation (HFS) of the presynaptic nerve terminal increased intracellular Ca²⁺ in MNTB neurons, a fraction of which (3.8%) is attributable to ASICs. Additionally, synaptic depression during HFS was increased in: (i) the presence of psalmotoxin-1, (ii) ASIC1a^-/- animals and (iii) in solutions with enhanced extracellular pH buffering capacity [89].

2.1.1.4. Synapses on Pyramidal Neurons of Anterior 
Cingulate Cortex

Postsynaptic ASICs also mediate a fraction of the EPSCs at glutamatergic synapses on pyramidal neurons of the anterior cingulate cortex [93]. Indeed, after blocking AMPARs and NMDARs, a small (~20 pA) component of the EPSC evoked by stimulating the deep layer of the cingulate cortex remained. This residual current was significantly inhibited by amiloride/PcTx1 and was mainly lacking in neurons of ASIC1a^−/− mice [93]. It has been demonstrated that in the anterior cingulate cortex ASIC1a is the major player in LTP and pain hypersensitivity; however, it is not required for LTD induction [93].

2.1.1.5. Synapses on Hippocampal CA1 Pyramidal Neurons

Early attempts to reveal any synaptic ASIC-currents in hippocampal neurons were not successful [41-94], presumably due to the small size and small relative contribution of synaptic ASIC-currents to total synaptic currents. Nevertheless, a recent study has demonstrated that ASIC1a contributes to the excitatory postsynaptic currents in CA1 pyramidal neurons [95]. In particular, it has been shown that a fraction of EPSC (about 6% of the total current) evoked by stimulation of the Schaffer collaterals fibers is blocked by a specific blocker of homomeric ASIC1a currents, PcTx-1. In agreement with the postsynaptic localization of ASIC1a, the paired-pulse ratio (PPR) of EPSCs, which is used to investigate the probability of neurotransmitter release, was unaffected following bath application of PcTx-1. It has been shown that ASIC1a is essential for long-term depression (LTD) evoked by low-frequency stimulation (LFS) and is involved in LTD induced by brief bath application of NMDA [95]. Additionally, a very interesting observation reported in this study [95] is a putative functional interaction between ASIC1a and NMDA receptors during LTD. In brief, LFS evokes LTD of glutamatergic PSCs; LTD is notably attenuated when ASIC1 is blocked. When AMPA-receptors are blocked (by CNQX) and NMDA-dependent component of glutamatergic PSCs is recorded, LFS stimulation also evokes LTD. However, the block of ASIC1 under these conditions evokes LTP instead of LTD [95].

The authors acknowledge that the mechanism of this phenomenon is currently unknown and suggest that it is due to a functional interaction between ASIC1a and NMDA receptors [95]. On the other hand, one of the differences between the series of experiments with ‘glutamatergic PSCs’ and ‘NMDA-dependent PSCs’, is the absence/presence of CNQX (AMPA/kainate receptor antagonist) throughout the experiments. Thus, it cannot be excluded that a functional interaction between ASIC1a and AMPA/kainate receptors occurs in the ‘glutamatergic PSCs’ series and this is blocked in the ‘NMDA-dependent PSCs’ series. Once this interaction is blocked, LTD becomes LTP.

In the conclusion of this subsection, it is worth mentioning that the absolute amplitudes of the synaptic currents mediated by ASICs are rather small in all of the above-mentioned synapses (Table 2). The relative contribution of the ASICs-mediated component to EPSCs is also modest (3 -10%). In spite of this, ASICs do play important roles in synaptic functions, such as regulation of short- and long-term plasticity. Moreover, in multiple cases described in the following section, ASICs apparently play some functional role even if synaptic ASIC-currents have not been detected. Additionally, it is worth mentioning that protonergic synaptic currents may be subject to modulation by endogenous substances [22, 24, 81, 96, 97]. Indeed, up-regulation of synaptic ASIC-mediated currents by neuromodulators, like histamine and natural products like lactate and spermine, has been recently reported [96]. Presumably, down-regulation of ASICs by GABA and glycine, as reported earlier [98], may also be relevant to the modulation of ASIC-mediated synaptic currents under physiological conditions.

Table 2.

ASICs as postsynaptic receptors activated by protons.

Brain Structure/Area, Type of Postsynaptic Neuron	Amplitude of Synaptic ASIC-current (pA)	Fraction of Total Synaptic Current (%)	Type of ASICs Involved in Synaptic Transmission	Refs.
Lateral amygdale, pyramidal neurons	7	6	ASIC1/ASIC2 heteromultimers	[58]
Nucleus accumbens, medium spiny neurons	30	5	ASIC1A/ASIC2A heteromeric	[66]
Medial nucleus of the trapezoid body	46	3	ASIC1a homomeric	[89]
Dorsal striatum, medium spiny neurons	18	10	ASIC1a homomeric	[92]
Anterior cingulate cortex, pyramidal neurons	17	10	ASIC1a homomeric	[93]
Hippocampal CA1, pyramidal neurons	21	6	ASIC1a homomeric	[95]

2.2. Modulatory/Regulatory Roles of ASICs at Glutamatergic Synapses and GABAergic Synapses

2.2.1. Glutamatergic Synapses

While synaptic ASIC-currents were not detected by Cho and Askwith [94], in the same study, this group reported definitive changes in excitatory synaptic transmission in ASIC1 knockout neurons. Namely: the AMPAR/NMDAR EPSC ratio was reduced in ASIC1 knockout neurons; the paired-pulse ratio (PPR) of AMPAR EPSC was reduced and the depression of AMPAR EPSCs during a short train of stimuli was greater in ASIC1 knockout neurons [94]. Further analysis revealed that presynaptic release probability was enhanced in ASIC1 knockout neurons. Since the above-mentioned changes of glutamatergic transmission were not mimicked by: a) the non-selective antagonist of ASICs amiloride, and b) increasing pH buffer capacity of the extracellular solution, the authors suggested that the effects were to due to an unconventional ASIC function independent of ion conduction [94].

In another study [62], no changes of baseline synaptic transmission were observed in hippocampal slices from Asic1a^−/− mice relative to wild-type controls. However, the facilitation of excitatory synaptic responses during HFS and long-term potentiation (LTP) were impaired in Asic1a−/− mice [62]. No changes in paired-pulse facilitation of synaptic responses were observed in Asic1a^−/− mice relative to wild-type. This is in agreement with a postsynaptic mechanism of ASICs involvement and predominant postsynaptic localization of ASICs, as observed in the same work [62].

It has been shown recently [99] that ASIC1a plays a role in the modulation of NMDAR EPSCs in hippocampal CA1 neurons. Indeed, both amiloride and PcTX1 substantially reduced the amplitude and slowed the activation and deactivation kinetics of the NMDAR EPSCs. On the other hand, no changes of PPR of NMDAR EPSCs were observed [99], implying that modulation was due to a postsynaptic mechanism. Since the increase in NMDAR current was still observed in the presence of blockers of CaMKII, PKA and PKC, it was concluded that the current was not mediated by these intracellular signaling pathways [99]. Further analysis revealed that ASIC1a modulates NMDA receptor function through targeting NR1/NR2A/NR2B heterotrimeric receptors and this is relevant for both physiological and pathological conditions [99].

Amplitude of synaptic ASIC currents and percentage of the total synaptic current are either values reported in original work (if provided) or estimates based on summary graphs/representative illustrations.

2.2.2. GABAergic Synapses

Many studies regarding the roles of ASICs in synaptic transmission have focused on glutamatergic synapses, as described in previous sections. However, synaptic cleft acidification also occurs at inhibitory GABAergic synapses [85]. Moreover, selective deletion of ASIC1a in GABAergic neurons also has important functional consequences [100]. Nevertheless, the potential role of ASICs at GABAergic synapses, to our knowledge, has been reported just in two studies.

In one of these studies, effects of three different blockers of ASICs, including amiloride, diminazene [101] and a novel potent orthosteric antagonist of ASICs named compound 5b [26] on evoked GABAergic postsynaptic currents (PSCs) were examined in hippocampal cell cultures [90]. The idea for the experimental design was as follows. If a fraction of postsynaptic current is due to activation of cation-selective ASICs, then PSCs are mediated by both cations and Cl^- anions. Given that the reversal potentials for cations and Cl^- are not the same, the different effects of ASIC blockers may be expected on synaptic currents recorded below and above the reversal potential of the PSCs [90]. Therefore, in the same cells, evoked GABAergic PSCs were recorded below their reversal potential as inward currents, and above the reversal potential, as outward ones. In both cases, however, the membrane potential was below the reversal potential for Na⁺. Thus a decrease of ‘inward PSCs’, and an increase of the ‘outward PSCs’ was expected if a fraction of PSC is mediated by cation-selective ASICs [90]. It has been reported that GABAergic PSCs, recorded below their reversal potential as inward currents, were suppressed by ~ 20% in the presence of compound 5b [90]. However, no increase in outward PSCs was observed, suggesting that ASICs do not mediate a substantial fraction of synaptic currents. In line with this suggestion, a fraction of postsynaptic current resistant to the blocker of GABA_A-receptors, bicuculline (20 μM), was not significantly affected by the ASICs blocker compound 5b [90].

The suppressing effect of compound 5b on ‘inward PSCs’ is attenuated in solutions with an enhanced concentration of extracellular pH buffer, implying that endogenous protons are involved and the effect is specific. The specificity of the effect is also supported by experiments with two other blockers (amiloride and diminazene). Similar to compound 5b, both of these blockers decreased ‘inward PSCs’ without significant effects on ‘outward PSCs’.

The main conclusions of this work [90] are: (i) that ASICs play a functional role at GABAergic synapses, and (ii) involvement of ASICs is at least partially postsynaptic but predominantly modulatory (see below). The following model can be proposed to explain the results.

1. Both GABA and protons are released upon stimulation of presynaptic GABAergic neurons and diffuse to the postsynaptic receptors.

2. The protons, which are smaller, reach postsynaptic receptors first and activate ASICs.

3. Activated ASICs interact with GABA_A receptors resulting in their up-regulation.

4. Both synaptic currents directly mediated by ASICs and the modulation of GABA_A receptors are suppressed by ASIC blockers. However, only the modulation is resolved, as more robust phenomenon [90].

Although hypothetical, this explanation is in agreement with: (i) acidification occurring at GABAergic synapses [85]; (ii) predominant postsynaptic localization of ASICs [62, 91, 58], (ii) functional crosstalk between ASICs and GABA_A-receptors, reported in isolated neurons [98], [102]; (iii) modest amplitude of synaptic ASIC-currents [58, 66, 95]; (iv) some other evidence [103, 104].

Interestingly, in terms of magnitude, the modulatory effect in GABAergic synapses [90] appears to be larger than the contribution of ASICs to synaptic currents in glutamatergic synapses. For example, the amplitude of the synaptic current modulated by ASICs at GABAergic synapses on hippocampal neurons is about 40 pA (~20% of the total current) [90], while the amplitude of synaptic current mediated by ASICs at hippocampal excitatory synapses is ~20 pA (6% of the total current) [95]. This may indicate that the functional role of ASICs at GABAergic synapses is even greater than that at glutamatergic synapses. Additionally, such a modulatory effect may be even more pronounced at inhibitory synapses on interneurons. Indeed, many studies showed that GABAergic inhibitory interneurons in the hippocampus have larger ASIC current densities than glutamatergic excitatory pyramidal neurons (PNs) [82, 94, 105] (but see [106] for conflicting evidence). This is especially true for oriens lacunosum-moleculare (O-LM) interneurons in which ASIC current density is almost seven times higher than that in PNs (0.75 pA/μm² versus 0.11 pA/μm²) [107]. Thus, it may be expected that the modulatory role of ASICs at GABAergic synapses on O-LM interneurons is particularly robust. An interesting aspect of the above-mentioned discussion is that the overall effect of ‘silencing’ ASICs on electrical activity in the neuronal network could be inhibitory. Disinhibition of excitatory neurons may be counterbalanced by more pronounced disinhibition of (inhibitory) interneurons.

It has been found in another study [108] that ASICs selectively control the frequency of spontaneous inhibitory synaptic activity in hippocampal neurons. In this study, effects of orthosteric ASIC antagonist compound 5b on excitatory (glutamatergic) and inhibitory (GABAergic) PSCs were studied in hippocampal slices and ‘synaptic bouton’ preparations (isolated cells with preserved synaptic boutons, see [109]). In hippocampal slices, the frequency of spontaneous excitatory postsynaptic currents (sEPSCs) was not changed by 5b administration [108]. In contrast, the frequency of spontaneous postsynaptic inhibitory currents (sIPSCs) was increased by 30% in the presence of compound 5b. The amplitude distribution of sIPSCs was not affected by compound 5b. Both the increased frequency of sIPSCs and the lack of change in the amplitude distribution suggest that the effect of compound 5b is due to a presynaptic mechanism. This suggestion is also supported by changes in the paired-pulse ratio of evoked PSCs elicited by compound 5b [108]. Effects of compound 5b on sIPSCs were reproducible in ‘synaptic bouton’ preparations, where ‘network’ effects are eliminated. In this preparation, the frequency of (sIPSCs), was increased by 74%, while the amplitude distribution of sIPSCs was not altered following administration of compound 5b [108]. Thus, the effect of compound 5b on sIPSCs is indeed presynaptic and is not due to changes in electrical activity in the hippocampal network.

Considering the correspondence between presynaptic inhibition and depolarization of the primary afferent terminals [110], it has been suggested that ASIC-mediated presynaptic depolarization is responsible for lower transmitter release under control conditions [108]. Thus, once ASICs are blocked, the transmitter release is enhanced.

The increase of frequency of sIPSCs evoked by compound 5b has been suggested as a mechanism underlying suppression of (i) epileptic discharges in a low Mg²⁺ model of epilepsy in hippocampal slices; and (ii) kainate-induced discharges in the hippocampus [108]. These results are discussed in more detail in the section related to pathologies. Here we will highlight the important role of GABAergic signaling in both physiological (e.g., theta and gamma oscillations or sharp wave-ripples) and pathological oscillations [111].

While it may seem that the results reported in [90] and [108] about GABAergic synapses on hippocampal neurons are conflicting, there are actually no direct contradictions. Indeed, the lack of amplitude change of the outward sIPSCs due to the administration of compound 5b [108] is in line with the lack of effect on outward evoked PSCs reported by [90]. Regarding the paired-pulse ratio of evoked PSCs: PPR was changed following administration of compound 5b [108]; however, no significant changes in PPR were evoked by this blocker according to [90]. Presumably, this difference can be explained by the absence [108] and presence [90] of CNQX (AMPA/kainate receptor blocker) in the extracellular solutions throughout the experiments. Indeed, it has been shown that CNQX by itself can presynaptically affect GABAergic PSCs [112, 113]. Moreover, it appears that CNQX may strongly affect the ASIC1-dependent plasticity of synaptic transmission (see [95] and discussion regarding this work in the previous subsection). Therefore, an alteration of an ASIC-dependent mechanism of regulation of synaptic transmission by CNQX in some studies is not very surprising.

Thus, the evidence is accumulating that ASICs play a functional role at GABAergic synapses, while both pre- [108] and postsynaptic [90] mechanisms are involved in ASIC-mediated regulation/modulation of GABAergic synaptic transmission. This functional role may be relevant to learning deficits observed in mice with selective deletion of ASIC1a in GABAergic neurons [100].

2.2.3. Role of ASICs in Use-dependent Synaptic Plasticity

The efficacy of synaptic transmission is not fixed but can be up- or down- regulated as a result of the previous activity of pre- and postsynaptic neurons. Depending on the pattern of this activity, the strength of a given synaptic connection may be increased or decreased. These use-dependent changes in synaptic strength are thought to underlie learning and memory [114]. It appears that ASICs play an important role in several forms of short- and long-term synaptic plasticity, as briefly described below.

2.2.4. Short-term Synaptic Plasticity

It has been reported that silencing ASICs affects a simple form of short-term plasticity termed paired-pulse depression. Namely, Cho and Askwith [94] delivered pairs of stimuli with short intervals (ranging from 20 to 400 ms in different experiments), recorded AMPAR EPSC and determined PPR. Compared with wild-type neurons, the PPRs (with interpulse intervals of 50, 100, and 200 ms) were significantly reduced in ASIC1 knockout neurons. Additionally, short-term synaptic depression during a train of repetitive stimuli with 50-ms intervals was reduced in ASIC1 knockout neurons compared with wild-type neurons [94]. ASIC-dependent changes of short-term synaptic plasticity during a train of stimuli applied at 100 Hz [62] and 300 Hz frequency [89] have also been reported. On the other hand, changes of PPR of glutamatergic EPSCs were not observed in other studies (20 and 50 ms interpulse interval [62]; 25, 40, 75 and 100 ms interpulse interval [93]).

2.2.5. Long-term Synaptic Plasticity

There is growing evidence that ASICs are involved in long-term synaptic plasticity, namely LTP and LTD (Table 3).

Table 3.

Involvement of ASICs in LTP and LTD.

Brain Structure/Area	Protocol of Induction	Form of Long-term Plasticity	Refs.
Hippocampus	HFS, theta burst stimulation	LTP	[62, 26, 115]
Hippocampus	PP-LFS	LTD	[116]
Hippocampus	LFS	LTD	[95]
Amygdala	HFS	LTP	[58, 100]
Insular cortex	LFS	LTD	[117]
Anterior cingulate cortex	HFS	LTP	[93]

HFS-high-frequency stimulation; LFS – low-frequency stimulation; PP-LFS - paired-pulse low frequency stimulation.

In particular, ASIC-dependant LTP has been reported in the hippocampus [62, 26,115], amygdala [58, 100] and anterior cingulate cortex [93]. Wemmie and colleagues [62] have reported that loss of ASIC impaired hippocampal long-term potentiation [62]. In this work, high-frequency stimulation (HFS) was used to induce LTP, while field excitatory postsynaptic potentials (fEPSPs) were recorded in hippocampal slices from wild-type and ASIC knockout mice. Immediately after HFS, slices from both genotypes showed an increase in fEPSP slope and amplitude, indicating that short-term potentiation occurred in both groups. In contrast, long-term potentiation was strikingly impaired in ASIC knockout mice [62]. Interestingly, in the presence of low Mg²⁺, both genotypes exhibited comparable LTP, suggesting that facilitation of NMDA receptor function may be sufficient to overcome the ASIC-dependent deficit in LTP [62].

In line with a key role of NMDA receptor-dependent synaptic plasticity in the CA1 region of the hippocampus in the acquisition and consolidation of spatial memory, ASIC knockout mice exhibit a mild deficit in spatial learning and memory [62]. On the other hand, disparate results regarding the importance of ASIC1a in hippocampal LTP and spatial learning have also been reported: robust LTP was observed in ASIC1a-null mice, while mice lacking ASIC1a had normal performance in hippocampus-dependent spatial memory [72]. These discrepancies may be due to the differences in genetic deletion approaches.

Another study has shown [26] that the orthosteric antagonist of ASICs, compound 5b, blocks LTP in CA3-CA1 but not in MF-CA3 hippocampal synapses. These findings support results reported by Wemmie and colleagues [62] and highlight the important role of ASIC1a channels in the NMDAR-dependent LTP [26]. This point was subsequently confirmed: genetic deletion or pharmacological blockade of ASIC1a greatly reduced (although did not abolish) the probability of long-term potentiation (LTP) induction by either single or repeated high-frequency stimulation or theta-burst stimulation in the CA1 region [115].

ASICs are also important players in LTP in the lateral amygdala [58]. To assess the role of ASICs in LTP in this structure, HFS was applied and excitatory postsynaptic potentials (EPSPs) were measured. Immediately after HFS, EPSPs were increased in slices from WT and ASIC1a^−/− mice. However, LTP was strikingly reduced in ASIC1a^−/− slices, decaying to baseline 15 min after HFS. Additionally, (in slices from wild-type mice) reducing pH buffering capacity of the extracellular solution strikingly enhanced LTP, while increasing buffering capacity diminished it [58]. This is in line with the idea that an increased proton concentration is required for this form of plasticity [58].

Interestingly, ASICs are differentially expressed within the amygdala neuronal population, and the extent of LTP at various glutamatergic synapses correlates with the level of ASIC expression in postsynaptic neurons [100]. Moreover, selective deletion of ASIC1a in GABAergic cells, including amygdala output neurons, eliminated LTP in these cells and reduced fear learning to the same extent as found when ASIC1a was selectively abolished in BLA glutamatergic neurons [100].

ASIC1a is also abundantly expressed in the anterior cingulate cortex, where it is critically involved in nociceptive hypersensitivity through modulation of the strength of excitatory synaptic transmission [93]. Indeed, it has been shown that disrupting ASIC1a expression/or function in this structure by either genetic deletion or pharmacological tools blocked cingulate LTP induction and this results in modulation of nociceptive hypersensitivity [93]. Mechanisms of ASIC1a involvement in these processes have been studied in much detail. In particular, it has been shown that LTP induction occurs through a PKCλ -dependent increase in GluA1 receptor membrane trafficking [93].

While the above mentioned studies have mostly focused on physiological LTP, ASICs are also involved in pathological forms of LTP, for instance, LTP induced by oxygen and glucose deprivation [118] or chronic pain [93]. In this regard, block of ASICs is considered as a strategy for neuroprotection [118] and treatment of chronic pain [119, 120]. The 
latter includes pain involving supraspinal mechanisms and ASIC1a [93].

Besides LTP, ASIC1a is also essential for LTD. So far, this has been demonstrated in the insular cortex [117] and hippocampus [116, 95]. Interestingly, apart from the suppression of NMDA-dependent LTD [117, 95], silencing of ASIC1a also inhibited metabotropic glutamate (mGlu) receptor-dependent LTD present in early adult (P30-40) but not in juvenile (P13-18) animals [116]. ASIC1a-mediated insular LTD is suggested as a mechanism in extinction learning [117], while the role of ASIC1a-mediated hippocampal LTD at the behavioral level (so far) is currently unclear.

In the conclusion of this section: ASICs are involved in several forms of short- and long-term synaptic plasticity in multiple brain regions; ASICs may play differential roles in synaptic plasticity depending on the pattern of neuronal activity, nature of the brain region and downstream signaling pathways. Therefore, it is not surprising that at the level of learning-related behavior, ASICs (especially ASIC1a) play a wide variety of roles. As ASICs are involved in pathological forms of synaptic plasticity and excitotoxicity [118], ASICs are considered as emerging drug targets for a number of pathologies [118-121].

2.3. Roles of ASICs in Learning and Behavior

The next question to be addressed in this review is the system effects of the ASICs in the brain. As we have mentioned before, ASICs are expressed in the hippocampus and amygdala, brain structures involved in complex behavior. Therefore, their activity may influence learning, emotional reactivity, and other forms of behavior. Several approaches are used to study the role of ASICs in complex behavior. The first is the genetic manipulation to create animals with either (i) globally disrupted [62] or overexpressed gene(s), [123] or (ii) altered expression restricted to certain brain structures [57, 66]. Another approach is acute pharmacological inhibition [57], or activation [122] of ASICs.

2.3.1. Anxiety-like Behavior

Exploration of a novel environment is one of the basic innate behavior typical for rodents. Open field (OF) and elevated plus maze (EPM, or its modification of the elevated zero maze EZM) are the two most common approaches in any battery of behavioral tests. Both of them allow assessment of the general level of locomotion (total distance traveled) and the level of anxiety-like behavior (tendency to escape the open space, i.e. to spent time near the walls in the open field, number of entries and total time spent in closed arms in the elevated plus maze). A set of other methods can be used to assess the anxiety level: forced swim test (FST, greater immobilization time in an inescapable water tank), tail suspension test (TST, greater immobilization time after suspending by the tail above the ground), dark-light test (DLT, shorter time to enter the dark zone = escape from light). The effects of ASICs activation or deactivation on anxiety-like behavior are summarized in Table 4.

Table 4.

The effects of ASICs activation or deactivation on anxiety-like behavior.

References	Animal	ASIC Treatment	Behavioral Effects
[39]	mice	ASIC1a-/-	EPM: no difference
[123]	mice	ASIC1a overexpression	EPM: no difference
[124]	mice	ASIC1a-/-	OF: no diff. in total distance           OF: ↑ central time
[57]	mice	ASIC1a-/-	FST: reduced immobility           TST: reduced immobility
-	-	PcTx-1 or A-317567 i.c.v. (block.)	FST: reduced immobility
-	-	ASIC1a-/- with further re-expression of ASIC1a in basolateral amygdala (BLA)	FST: no diff. in immobility time compared to controls (functional restoration)
[125]	mice	PcTx-1 i.c.v. (block.)           A-317567 i.p.           (block.)           amiloride i.p.(activ.)	EZM: no difference
[40]	mice	ASIC1a-/-           ASIC2-/-           ASIC12-/-	OF: no difference           EPM: no difference           FST: reduced immobility
[122]	rats	PcTx-1 injection to BLA (block.)	OF: no diff. in total distance           OF: ↓ central time
-	-	Ammonium injection to BLA (activ.)	OF: no diff. in total distance           OF: ↑ central time           DLT: ↑ latency to enter the dark compartment
[92]	mice	ASIC1a-/-	OF: no diff. in total distance
[126]	rats	Compound 5b i.p. (block.)	OF ↓ total distance           OF no diff. in central time

* i.c.v. - intracerebroventricularly; i.p. – intraperitoneally.

Overall, genetic or pharmacological brain-wide manipulation of ASIC1/2 (both positive and negative) does not affect basic locomotion in simple exploratory tests (OF, EPM/EZM). The exception is our work [126], but it can be explained by relatively high doses of the ASIC1a blocker we used. The behavioral data of [124, 40] support the idea of the anxiolytic effect of ASICs inactivation, and this has additional confirmation from several stress-inducing tests [125]. In contrast, Pidoplichko et al. found that local activation of ASICs in the basolateral amygdala with the injection of ammonium leads to an increase in anxiety-like behavior in rats, and the inhibition of those channels with the psalmotoxin has the opposite effect [122]. Changes in the activity of ASICs in other than amygdala brain regions during brain-wide application of pharmacological agents is the most probable explanation for this discrepancy. In basolateral amygdala ASICs contribute to both excitatory and inhibitory postsynaptic potentials producing net inhibitory effect [123]. But in the hippocampus, the net inhibitory effect can be elicited with the inactivation of ASICs [108], which manifests in the general reduction of locomotor activity and the corresponding decrease in the peak-frequency of the hippocampal theta-band [126].

2.3.2. Locomotion and Motor Learning

It is known that ASIC1a is widely expressed in striatal neurons [67], which are involved in motor control and procedural learning. As mentioned above, general locomotor activity in the open field test and elevated plus-maze test is unaffected by the modulation of ASICs. Early experiments showed similar accelerating rotarod performance in ASIC1a^-/- and control mice [62], and this was confirmed later [92]. However, increased task difficulty (incremental rotarod, balance beam test) led to poorer performance of ASIC1a^-/- mice compared to WT. Prolonged training (3-5 days) improved the coordination of movements, diminishing the difference between KO and WT mice. Re-expression of ASIC1a in the dorsal striatal neurons rescued the defects in motor learning [92]. Such changes caused by the global loss of ASIC1a were largely mediated by reduced activation (phosphorylation) of Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) and extracellular signal-regulated protein kinases (ERKs) [92]. Acute inhibition of ASIC1a with PcTx-1 injection into dorsal striatum did not affect performance in the balance beam test but decreased the learning efficiency during prolonged training. Together, these data show the role of ASIC1a in procedural learning and memory.

2.3.3. Fear and Negative Emotions

Amygdala is the essential structure involved in emotional reactions, especially negative ones [127] and is also one of the brain regions with high ASICs expression [39]. That is

why the role of ASICs in fear reactions, learning, and extinction has been extensively studied during the last decades. To study innate fear, trimethylthiazoline (TMT) predator odor, a component of fox feces, is used [128]. It was shown that the global loss of ASIC1a [124, 40], or ASIC2, or both [40], partial loss of ASIC1a in most but not all neurons [129], and acute channel inactivation by PcTx-1 [124], decrease the freezing response to TMT administration. This effect is not rescued by restoring the ASIC1a in the basolateral amygdala, and this confirms the idea that innate fear processing depends more on the bed nucleus of stria terminalis than the BLA [130]. Another form of unconditioned fear – freezing in response to a painful stimulus (electric shock) – also was reduced in ASIC1a-null mice, and this effect remained after re-expression of ASIC1a into BLA [130].

The role of ASICs in acquired fear is more complicated. Experiments revealed cue and context fear conditioning deficit in global ASIC1a^-/- [39, 40], ASIC2^-/-, and ASIC1/2^-/- mutants [40], as well as the opposite pattern in animals with globally overexpressed ASIC1a protein [123]. The restoration of ASIC1a expression in BLA of ASIC1a^-/- mice rescued context-dependent fear memory, but not the stimulus-driven response as mentioned above [130]. Further experiments showed that such fear memory formation is dependent on the LTP at cortico-BLA synapses and intra-amygdalar synapses on output GABA-ergic neurons (medial part of the central amygdala), which, in turn, is dependent on the ASIC1a functioning [100].

Beyond learning, the next important question is the extinction of fear, which is practically related to the correction of post-traumatic stress disorder. In this regard, the most crucial structure is the ventral hippocampus. The efficiency of extinction was affected by local genetic (injected virus vector) or acute pharmacological (injection of PcTx-1) inactivation of ASIC1a in the ventral hippocampus (vHPC), but not in the dorsal hippocampus, medial prefrontal cortex (mPFC) or basolateral amygdala [131]. The mice with inhibited vHPC ASIC1a demonstrated impaired extinction (significantly longer freezing in the retrieval phase of the experiment compared with controls), and vice versa – the overexpression of the ASIC1a gene in this region led to the promotion of the extinction [131]. The authors proposed a model, in which postsynaptic ASIC1a activation enhances the expression of brain-derived neurotrophic factor (BDNF) in vHPC, which alters the functioning of NMDA receptors in vPHC-mPFC synapses.

ASICs also play a role in taste aversion conditioning. Here the positive taste stimulus (saccharine) is combined with the induced negative state (i.p. injection of LiCl). The disruption of ASIC1a (both genetic and pharmacological) also impairs extinction without altering the learning process [117]. This can be restored by re-expression of ASIC1a in the insular cortex, whose projections to BLA are thought to be important for taste-based learning [132]. Interestingly, ASIC1a is necessary for LTD but not LTP in the insular cortex [117].

A special type of fear is CO₂-induced panic. This is the only fear that people with a damaged amygdala can feel [133]. An increase in inhaled CO₂ concentration leads to a substantial reduction of pH in the amygdala [134]. Freezing is the typical behavioural response to hypercarbia, which becomes prominent at approximately 10% of CO₂ in the air. This type of behavior is impaired in ASIC1a-, ASIC2-, and ASIC1,2-null mice as well as in ASIC1a^+/+ mice after 
pharmacological blockage of the channels (PcTx-1 and 
A-317567, i.c.v.) [40, 134]. Remarkably, the ventilation response to hypercarbia was not altered due to ASIC1a inhibition. Also, the lesion of the amygdala in mice led to a decrease in freezing but an increase in jumping behavior [129], revealing the dual role of this structure in the CO₂ response. Altogether, these results confirm the anxiolytic action of ASIC1a inhibition [125].

2.3.4. Positive Reinforcement Learning

The effect of ASICs in conditioning may depend on the order of conditioned (CS) and unconditioned (US) stimuli. In classical conditioning, when CS (sound + light) fully precedes US (“Ensure” based food) ASIC1a^+/+ animals demonstrate better ability to learn then ASIC1a^-/- [135]. This difference disappears in cases where CS and US partially overlap. Surprisingly, when CS and US start synchronously (CS loses part of its informativeness since the US is presented at the same time), ASIC1a^-/- animals perform better than controls. In cases of a fully random order of presentation of stimuli, this effect is absent, so it is likely to be due to some kind of conditioning. These data point to an intriguing but yet unknown role of ASIC1a in the probability calculation feature of brain reward systems.

2.3.5. Spatial Memory

Early work has shown the difference in Morris water maze performance between ASIC1a^+/+ and ASIC1a^-/- mice [62]. This becomes prominent starting from day 3 of training and can be normalized by more intense training. However, later experiments did not confirm such difference, showing the same performance of control and ASIC1a-null mice [136]. Another paper connects ASICs with spatial memory, in which authors found that dietary walnut oil improved performance of rats in the Morris water maze [137]. Also, adding this substance to the hippocampal neurons culture medium leads to an increase in the expression of ASIC2a and ASIC4 genes. Authors conclude that intake of walnut oil might help to prevent cognitive decline, but the big question is if any component of this oil can get unaltered through the brain-blood barrier.

2.3.6. Cocaine-seeking Behavior

Nucleus accumbens (NAc) is another ASICs-reach brain structure, which is involved in addiction behavior, particularly associated with cocaine [138]. An acute cocaine-induced dose-dependent increase in locomotor activity was observed in control, ASIC1a^-/-, and ASIC2^-/- mice [139]. This increase was lowest in ASIC1a-null animals at all drug doses. The sensitization effect after five consecutive days of cocaine administration followed by a 2-week withdrawal period was most prominent in ASIC1a^-/- mice. In the case of ASIC2 knockout, the sensitization effect was observed during a short period immediately after cocaine injection. Local disruption of ASIC1a in NAc increased cocaine-conditioned place preference and restoring this gene with virus injection reduced the time spent on the cocaine side of the chamber [66]. These effects were accompanied by changes in the glutamate system and dendritic spine density in NAc neurons. Interestingly and importantly, ASIC1a overexpression led to decreased cocaine intake in the self-administration test in rats, which is of interest for developing a therapeutic strategy to treat drug addictions [66]. A recent paper shows increased cocaine-seeking behavior in rats with overexpressed ASIC1a in NAc [140], which highlights the necessity for further research in this field.

ASICs may play a role in the organization of behavior through other mechanisms. There is evidence supporting the existence and activity of ASICs in the anterior hypophysis in mice [58], which may connect their activity to hormonal regulation of homeostasis and behavior. Also, acute stress affects ASICs current in hippocampal neurons and increases learning and memory via glucocorticoid receptors [141]. Overall, we can conclude that ASICs in the brain are involved in many important processes that have behavioral manifestations, including different types of learning.

2.4. Other Roles of ASICs in the Brain

It is important to note that many important findings regarding the roles of ASICs in the brain are just outside the scope of this mini-review. Nevertheless, we would like to mention a very interesting recent finding for the role of ASIC1a as a cerebral pH sensor responsible for the regulation of microvascular tone [142]. According to this study, the activation of ASIC1a, particularly in neurons, is critical for CO₂-induced production of nitric oxide and vasodilatation [142]. Thus, ASIC1a emerges as a major regulator of microvascular tone [142]. This is certainly very important for both physiological and pathological processes in the brain and requires further investigation [143].

3. PATHOLOGICAL ROLES OF ASICS IN THE BRAIN

A number of neurological disorders involve prolonged acidosis, arising from several possible sources, including stroke, ischemia, inflammation, etc. [144]. This results in acidosis-induced injury, which may eventually kill neurons and worsen the course of the disease. There is growing evidence that ASICs mediate acid-induced toxicity in the CNS and that inhibiting ASICs has protective effects [144, 145]. The pathological role of ASIC1a in stroke is well established and extensively reviewed ([146-149]). As was mentioned before, homomeric ASIC1a channels conduct Na⁺ as well as Ca²⁺ ions. Blocking ASIC1a channels greatly reduces infarct volume, and potential neuroprotective drugs based on this effect will thus have a prolonged therapeutic window [147].

In this section, we will briefly review evidence regarding the potential protective roles of ASICs inhibition in some other neurological disorders.

3.1. Epilepsy

About 1% of the human population suffers from chronic epilepsy disorder and 30-40% of patients develop drug-resistant epilepsy. Finding new targets for pharmacological intervention to treat patients with epilepsy is strongly needed. Epilepsy is a neurological disorder characterized by unprovoked seizures, about 40% of cases are idiopathic. An epileptic seizure is a pathological electrical signal in the brain that causes changes in the behavior, feelings, and movements. The seizure-like activity can be detected by recording electrical signals from the brain in vivo as well as in vitro models [150].

The intensive electrical activity of neurons during epileptic seizures leads to a significant decrease in pH in the brain tissue, suggesting a possible role of ASICs in the regulation of epileptic discharges. Some authors propose that acidification following increased neuronal signaling activates ASIC and diminishes seizure severity. It is known that ASIC density on hippocampal inhibitory neurons is larger than on principal neurons. Authors of the proposed mechanism speculate that activation of ASIC increases inhibitory neurons’ tone, and as a result decreases overall neuronal activity [105].

Other studies demonstrate that inhibition of ASICs activity reduces epileptic manifestation in vitro models of epilepsy. Perfusion of hippocampal slices with amiloride or PcTX1 reduces epileptic discharges evoked by electrical stimulation or removal of extracellular Mg²⁺. Hippocampal slices obtained from ASIC1a knockout mice demonstrate decreased seizure susceptibility in low Mg²⁺ and electric stimulation models of epilepsy [151].

Application of novel ASIC1 antagonist compound 5b to rat hippocampal slices during epileptic activity in low Mg²⁺ extracellular solution significantly, and reversibly, reduced frequency of seizure discharges [108]. In an in vivo model of epilepsy, selective ASICs antagonists reduced seizure-like activity evoked by injection of kainic acid in the amygdala or hippocampus [108, 151].

An increasing rate of excitatory signaling in the brain can lead to epileptic discharges in the neuronal network. Pharmacological agents that increase the activity of membrane inhibitory channels increase neuronal tissue resistance to seizures. Perfusion of rat hippocampal slices with selective ASIC1 antagonist 5b significantly increases the frequency of GABAergic, but not spontaneous glutamatergic currents. It was suggested that the locus of ASICs' effect on GABA release probability is presynaptic. Blockade of presynaptic ASICs increased interneuron signaling that resulted in a decrease in overall neuronal activity. Also, the downregulation of ASIC1a expression in astrocytes reduced spontaneous seizures in rats [52].

However, another study demonstrated that ASIC3 inhibitor APETx2 increased seizure susceptibility by upregulation of the N-methyl-D-aspartate subtype of glutamate receptors. The authors reported that ASIC3 expression was increased in the hippocampus of pilocarpine-induced seizure rats [152].

Altered expression of ASIC isoforms in the cortex after status epilepticus was reported. While the expression of ASIC1a was decreased in the acute phase, the expression of ASIC2a was upregulated. Rats with ASIC2a knockdown have reduced dendritic length and spine density in primary neurons. The authors propose that the upregulation of ASIC2a after seizure contributes to dendritic lengthening [153].

The role of ASICs in the regulation of epileptic activity was demonstrated in a study devoted to the effect of ketone bodies on the activation of ASICs. The ketogenic diet is effective in managing pharmacoresistant epilepsy, but the mechanism of anticonvulsive effects of ketone bodies remains unclear. It was demonstrated that ketone bodies effectively inhibit the opening of ASICs [106].

WIN55,212-2 cannabinoid receptor agonist dose-dependently inhibits ASICs in rat dorsal root ganglion. The effect almost completely depends on CB1, but not the CB2 receptor. Whether this analgesic mechanism is relevant to the antiepileptic properties of WIN55,212-2 requires future research [154].

3.2. Anxiety and Depression

The amygdala is abundant in ASIC receptors. The high density of ASIC receptors in this structure suggests their profound role in fear and anxiety. Studies on human subjects demonstrate that panic disorder correlates with ASIC1 polymorphism [155, 156]. Some authors report that activation of ASIC1a in the basolateral amygdala decreases anxiety-like behavior, while inhibition of ASIC1a increases the level of anxiety in rats [122].

Other studies demonstrate that activation of ASIC1a receptors significantly increases fear and anxiety. Data suggest anxiolytic properties of ASIC antagonists, for example, acute administration of PcTX-1, A-317567, or amiloride prevented stress-induced elevation in core body temperature in the stress-induced hyperthermia model. In addition, the authors report that injection of A-317567 (100µM) significantly increases the level of GABA, but not glutamate in the amygdala [125, 157, 158].

Disrupting the ASIC1a gene decreased fear in the open field test, reduced acoustic startle, and decreased trimethylthiazoline (predator odor) evoked freezing [124, 159]. Also, selective ASIC1 blocker compound 5b decreased hippocampal theta frequency in a manner similar to classical anxiolytics such as benzodiazepine and barbiturates [126].

ASIC1a is proposed as a target to combat depression. Mice with genetically disrupted ASIC1a demonstrate reduced depression-like behavior in the forced swim test, tail suspension test, and unpredictable mild stress. Restoring ASIC1a in knock-out mice with viral vector reverses the forced swim test effect. Pharmacological inhibition of ASIC1a also produces antidepressant-like effects in the forced swim and was independent of commonly used antidepressants [57].

Experimental data suggest that ASIC receptors could be involved in mechanisms that underlie psychiatric disorders and that modulation of ASIC receptors could improve the treatment of patients with anxiety disorders and depression.

3.3. Alzheimer's Disease

Alzheimer's disease (AD) is a progressive neurodegenerative disorder, the most common form of dementia, accounting for 60–70% of all cases [160]. In spite of many attempts, currently, there are still only symptomatic treatments available for this disease; no drugs capable of arresting or reversing it have so far been identified [160, 161]. Although the pathological accumulation of β-amyloids is a major histopathological hallmark of AD, the question of whether this is the primary cause of the disease is still debated. In any case, it is widely accepted that the progression of AD occurs due to multiple factors. There is evidence that activation of ASICs may be one of these factors (reviewed recently [121]).

Pharmacoepidemiology and observational research play an important role in the identification of potential therapeutic targets for disease-modifying treatments. In this context, it is worth mentioning that: (i) amiloride, a non-specific blocker of ASICs, is a well-known potassium-sparing diuretic used in medicine; (ii) use of diuretics (including amiloride) is associated with reduced risk of AD [162-163]. Additionally, while all antihypertensive drugs are associated with reduced risk of AD, potassium-sparing diuretics (including amiloride) are associated with the greatest reduction in AD risk [162]. Moreover, “analysis with a fully examined subsample controlling for blood pressure measurements did not substantially change” these findings [162]. While these results do not necessarily indicate that the observed association is due to amiloride action on ASICs, they do support this possibility. Additionally, it may be not coincidental that: curcuminoids, considered as a treatment for AD [161], block ASICs [164]. Furthermore, a selective blocker of ASIC1a, psalmotoxin-1, rescued the DHPG-LTD facilitation associated with genetic and non-genetic models of AD [165].

In this context, a comparison between potential protective effects of (i) potassium-sparing diuretic blocking ASICs (amiloride) and (ii) diuretics which do not directly affect ASICs, against AD may give valuable clues. Such studies may provide an answer to the question: does the block of ASICs reduce the risk of AD in humans? Importantly, since the information required for this comparison has already been collected, the comparison should proceed without delay.

3.4. Neonatal Hyperbilirubinemia

Neonatal hyperbilirubinemia has been known to damage neural function and morphology [166-168]. Accumulation of bilirubin in the CNS results in neurotoxicity in various brain regions, including the ventral cochlear nucleus, vestibular nuclei, cerebellum, and hippocampus [169]. Bilirubin-induced neurotoxicity is widely accepted as a leading cause of hearing loss, balance, motor control, and cognitive deficits in children with severe clinical hyperbilirubinemia, and is often accompanied by such factors as prematurity, infection, sepsis, hypoxia, ischemia, and acidosis [170, 171]. All of them may increase the permeability of the blood-brain barrier to bilirubin. The acidosis may promote the accumulation of bilirubin in the developing brain where developing neurons are vulnerable to its toxicity [172-174] and directly activates ASICs [144-145] in hypoxic and ischemic conditions.

In our recent paper, we found that neonates experiencing hyperbilirubinemia coinciding with acidosis have an increased concentration of lactate dehydrogenase in cerebrospinal fluid than patients with either condition alone, and this is the evidence of synergistic neurotoxicity [174, 175]. We demonstrated that bilirubin exhibited marginal neurotoxicity on its own, but potentiated the currents mediated by ASIC1a channels in an acidic environment via Ca²⁺-dependent intracellular signaling and augmented neuronal excitability, Ca²⁺ overload, and cell death [175]. Consistent with these results in vitro, neonatal conditioning with concurrent hyperbilirubinemia and acidosis primed long-term impairment of sensory and cognitive deficits in vivo in mice. These findings suggest the role of ASICs as novel therapeutic targets for the treatment of neonatal hyperbilirubinemia.

CONCLUSION

ASICs came to fruition in 1980. The concept of a highly selective Na⁺ current gated by specific receptors for protons was not readily accepted. It took 16 years to get the first representatives of these channels to be cloned. With this, a new era in ASIC investigation was started. “The receptor for protons” became ASICs comprising a family of receptor/
channels ubiquitous for the mammalian nervous system, both peripheral and central. This distribution makes these channels important players in arranging the brain's neuronal ensembles. Now ASICs are known as a family of at least eight Na⁺ channels encoded by five different genes. The proteins encoded by these genes represent subunits that form homo- and heterotrimers with distinct biophysical and pharmacological properties. Virtually all known ASIC subunits are present in the mammalian brain with the major representation of ASIC1a, 2a, and 2b. ASIC1a subunit is most sensitive to protons and conducts both Na and Ca ions. It is quite intriguing that some ASIC subunits, despite their name, e.g., ASIC4, ASIC5, and ASIC2b, do not gate ion currents when protonated.

Under physiological conditions, the rapid acidification occurring during synaptic vesicle release activates ASICs on both pre- and postsynaptic neurons. In the latter case, a fraction of postsynaptic current is mediated by ASICs with protons acting as co-transmitters. Additionally, activation of ASICs modulates the synaptic strength by affecting transmitter release and/or sensitivity of postsynaptic receptors and thus, contributes to use-dependent synaptic plasticity. There is growing evidence indicating that ASICs are involved in the regulation of synaptic function in multiple types of synapses, including both glutamatergic and GABAergic in several brain structures. ASICs play differential roles in synaptic plasticity depending on the pattern of neuronal activity, particularities of the structure of a brain region, and downstream signaling pathways.

The involvement of ASICs in synaptic plasticity underlies their region-specific effects on behavioral learning. ASICs (mainly ASIC1a) in striatal neurons are necessary for motor learning; in the basolateral amygdala — for conditioned fear-related freezing; in the ventral hippocampus – for the extinction of such fear; and in the insula — for the extinction of taste-related aversion. Also, ASIC1a is involved in CO₂-related fear (amygdala), and other forms of innate fear reactions (bed nucleus of stria terminalis). In the nucleus accumbens, ASICs alter addictive behavior. Our understanding of the role of these channels in complex behavior is still far from complete and requires further research.

Apart from the physiological roles of brain ASICs, they contribute to pathological states like epilepsy or psychiatric disorders and their correction. ASIC1 is a prominent target for anticonvulsive pharmacological intervention. Experimental results suggest that modulators of ASICs could improve the treatment of patients with anxiety disorders and depression. There is also emerging evidence indicating that blocking of ASICs may reduce the risks of Alzheimer's disease.

LIST OF ABBREVIATIONS

AMPA

α-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

AMPARs

AMPA Receptors

ASICs

Acid-sensing Ion Channels

CaMKII

of Ca²⁺/Calmodulin-dependent protein kinase II

CNQX

6-Cyano-7-nitroquinoxaline-2,3-dione;

(D-AP5)

D-2-amino-5-phosphonovaleric acid;

DLT

Dark-Light Test

EPM

Elevated Plus Maze (its alteration elevated zero maze EZM)

EPSCs

Excitatory Postsynaptic Currents

FST

Forced Swim Test

GABA

Gamma-aminobutyric Acid

HFS

High-frequency Stimulation

i.c.v

Intracerebroventricularly

i.p.

Intraperitoneally

LTD

Long-term Depression

LTP

Long-term Potentiation

NMDA

N-methyl-d-aspartic Acid

NMDARs

NMDA Receptors

OFT

Open Field Test

PcTx

Psalmotoxin-1

PKA

Protein Kinase A

PKC

Protein Kinase C

TST

Tail Suspension Test

CONSENT FOR PUBLICATION

Not applicable.

FUNDING

None.

CONFLICT OF INTEREST

The authors declare no conflict of interest, financial or otherwise.