ASIC3 Channels Integrate Agmatine and Multiple Inflammatory Signals through the Nonproton Ligand Sensing Domain

The fact that AGM directly activates ASIC3 channels [10] prompted us to look for the effects of polyamines and other arginine metabolites [11] (Figure 1A). Conventional whole-cell patch clamp recordings were performed in Chinese Hamster Ovary (CHO) cell lines transiently expressing ASIC3 tagged with GFP to measure the functional activation of ASIC3 channels under voltage clamp conditions. At a concentration of 1 mM, only AGM and its analog ARC (Figure 1B), but not polyamines (including spermine, spermidine, and putrescine), nor L-arginine, nor L-ornithine, were able to activate the ASIC3 channel at pH 7.4. The resistance of ASIC3 to polyamines differs from the modulation of ASIC1a by spermine [29, 30] and the modulation of TRPV1 by polyamines [24], arguing for multifunctional roles of arginine metabolites in inflammatory responses [40]. Therefore, as two major sensors for inflammatory pain, ASIC3 and TRPV1 sense different arginine metabolites (AGM of ASIC3 vs. polyamines of TRPV1, respectively).

AGM is widely and unevenly distributed in mammalian tissues [41]. Observed first in the rat brain, AGM was shown later that its concentration in the brain (2.40 ng/g) is lower than in others, such as stomach (71.00 ng/g), intestine, spleen and liver (5.63 ng/g) [42]. The relative low concentration of AGM in normal tissues together with the relative low potency of AGM on ASIC3 channels (3.3 ± 0.3% of the GMQ's response, Figure 1B) argue for the negligible contribution of AGM under physiological conditions. However, AGM is detectable in human plasma and higher concentrations have been observed in depressed patients [43]. Similarly, levels of peripheral AGM and polyamines are elevated during infection, trauma, and cancer [31, 32]. Interestingly, AGM evokes ASIC3-dependent pain-behavior in mice [10], suggesting that AGM-ASIC3 interaction may become functionally relevant under pathological conditions.

Pain conditions are associated with multiple inflammatory signals (mild acidosis, hyperosmolarity, increased arachidonic acid, or reduced extracellular Ca²⁺). To understand the functional relevance of AGM-ASIC3 interaction, we examined the pH dependence of AGM action by applying graded pH to the CHO cell expressing ASIC3 channels, in the presence or absence of AGM (Figure 2A). We found that when the typical biphasic ASIC3 responses were evoked by a pH reduction, AGM dramatically enhanced the sustained component (Figure 2A, C), without altering the peak component (Figure 2A, B). The effect was most evident under mild acidosis conditions (pH 7.2~6.8). This enhancement could be simply the summation of two independent currents induced by AGM and mild acidosis, respectively. To address this issue, we explored the interaction between AGM and pH 7.0 more specifically (Figure 2D-F). We found that pH 7.0 significantly potentiated AGM response and the potentiation was always more than additive with pH reduction regardless the sequence of agonist application (Figure 2D, E), a characteristic that is opposite to the accelerated ASIC desensitization caused by pretreatment with acidic solution [36]. This potentiation occurred only under the simultaneous presence of two ligands (i.e., H⁺ and AGM) (Figure 2D, E, panels I and III), suggesting coincident detection of proton and nonproton ligands by ASIC3 channels in vivo. Remarkably, a significant potentiation was observed even when AGM was applied at lower concentrations (~100 μM, Figure 2F). The inability of AGM pretreatment to enhance ASIC3 response to mild acidosis (Figure 2D, E, panel II) suggests a novel mechanism underlying the observed synergy between H⁺ and AGM that differs from the allosteric effect of tarantula toxin psalmotoxin 1 (PcTX1) on ASIC1 channels [44, 45].

It is well known that peripheral pH falls to < 7 during inflammation, infection, ischemia, hematomas, and exercise [1]. Moreover, such acidosis is well recognized to activate nociceptors and to produce pain in humans that can be attenuated by the ENaC/DEG inhibitor amiloride [46‐48]. Additionally, inflammatory mediators, such as nerve growth factor (NGF), serotonin (5-HT), interleukin-1, bradykinin, and brain-derived neurotrophic factor (BDNF) can stimulate ASIC3 transcription, which perhaps contributes to the pain-enhancing effects of these mediators [49, 50]. Thus, ASIC3 channels seem to act as a major inflammatory pain integrator [8, 9]. Considering that both AGM production [31, 32] and ASIC3 expression [51, 52] are increased during inflammation, the positive synergy between H⁺ and AGM in activating ASIC3 channels adds a new level of complexity to the molecular events that can lead to inflammatory pain. The dramatic enhancement under pH 7.2-6.8 (Figure 2C) is reminiscent of a previous observation reporting sustained 'window' current through ASIC3 channels at modest pH changes presumably contributing to myocardial ischemia [36]. Whether AGM regulates cardiac pain-sensing [36] and other forms of muscle pain [9, 52, 53] awaits further investigations.

In inflamed or injured tissues, multiple mediators meet in the interstitial fluid and form an inflammatory exudate, the content of which is acidic [34] and hyperosmotic [37]. Previous studies have shown that hyperosmolarity increases neuronal excitability in DRG neurons [8], affecting preferentially the sustained component of ASIC3 currents. These previous studies promoted us to examine the synergy among hyperosmolarity, acidosis, and AGM. ASIC3-expressing CHO cells were exposed to AGM and mild acidosis in the absence or presence of hyperosmolarity (Figure 3). The hyperosmolarity (600 mosmol kg^-1 with mannitol) itself did not induce any detectable current (Figure 3A) but significantly potentiated the ASIC3 currents evoked by a pH reduction to 7.0 (Figure 3), an effect previously observed in rat DRG neurons or an ASIC3 expressing F-11 DRG cell line [8]. Similarly, the current induced by AGM was markedly enhanced by hyperosmolarity (Figure 3). Interestingly, AGM caused further increase over the current induced by the combination of pH 7.0 and hyperosmolarity (Figure 3), suggesting that AGM, mild acidosis, and hyperosmolarity act synergically to facilitate ASIC3 opening, which may explain the enhanced sensory neuronal excitability under conditions of inflammation [8] or cardiac ischemia [36].

Next, we tested arachidonic acid (AA), a pro-inflammatory and ischemic factor which enhances ASIC currents induced by acid and increases neuronal excitability [8, 54, 55]. At the physiological normal pH, AA induced negligible currents from ASIC3-expressing CHO cells (Figure 4A). As expected, AA significantly potentiated AGM (1 mM)-induced currents (Figure 4). The relatively slow developing kinetics of the AGM current following addition of AA (Figure 4A) presumably reflects the slow onset of AA effect which requires several minutes to be fully established [8]. Therefore, AGM not only activates ASIC3 by itself at normal pH (Figure 1; Ref. [10]), but also exerts positive cooperative effect when administrated with other inflammatory factors such as mild acidosis, hyperosmolarity, and AA, further strengthening the notion that ASIC3 channels act as a multiple sensor to integrate diverse signals present in the pathophysiological environment [8, 9].

We also investigated the sensitivity of AGM-induced currents to alterations of extracellular Ca²⁺, which has marked effect on the GMQ response [10]. As shown in Figure 5A, B, the presence of 10 mM extracellular Ca²⁺ completely abolished the AGM currents, whereas reducing Ca²⁺ significantly potentiated AGM-induced currents. Low Ca²⁺ itself evoked a significant inward current (Figure 5A, left panel), consistent with a previous report [56]. Thus, similar to GMQ [10], AGM-ASIC3 interaction is highly sensitive to altered extracellular Ca²⁺. The extracellular Ca²⁺ concentration can decrease from a resting value of around 1.2-1.8 mM to values as low as 0.08 mM under certain conditions [38], suggesting that the signaling cascade induced by AGM-ASIC3 interaction might be markedly amplified under such conditions. Moreover, lactate produced by anaerobic metabolism reduces extracellular Ca²⁺ concentration and results in the enhancement of the acid-induced ASIC currents in ischemia-sensing neurons [39]. Likewise, AGM-evoked currents were increased about 3-folds when lactate and AGM were co-applied (Figure 5C, D). These results suggest that ASIC3 channels may be gated in vivo by the combined actions of reduced extracellular Ca²⁺, mild acidosis, and AGM through synergic interactions reported here.

It is, however, possible that AGM-induced ASIC3 channel activation was through the chelation of extracellular Ca²⁺ as observed with lactate [39, 56] at neutral pH. To clarify this possibility, we recorded ASIC3 response to AGM in Ca²⁺-free external solution. We found that AGM activated ASIC3 channels regardless the presence or absence of extracellular Ca²⁺ (data not shown). That AGM activates ASIC3 channels independent of Ca²⁺ chelation is consistent with the notion that AGM modulates ASIC3 activation via novel mechanisms (Figure 2D, E).

Next we asked how AGM activates ASIC3 channels in a manner similar to GMQ, given the difference in their structural flexibility (linear AGM vs. circular GMQ with a heterocycle, Figure 1A) [10]. In a previous study, we have shown that the nonproton ligand sensing domain plays a critical role in mediating AGM and GMQ effects on ASIC3 [10]. While its critical role for GMQ was supported by the fact that covalently linking circular GMQ or TNB to C79 activated ASIC3^E79C channels (with the GMQ-dimer or DTNB treatment) [10], the role of the same site for the linear AGM was not established. For comparison, we tested 2-aminoethyl-methanethiosulfonate (MTSEA), a linear thiol-reactive compound on ASIC3^E79C channels, in which the residue Glu79 was replaced by a cysteine, thus mimicking AGM-E79 interaction (Figure 6A). Bath application of 0.2 mM MTSEA persistently activated ASIC3^E79C channels at pH 7.4 (Figure 6B, D). By contrast, MTSEA (0.2 mM) was ineffective in CHO cells expressing wild-type (WT, data not shown) or ASIC3^E423C channels (Figure 6B). Interestingly, 2-(trimethylammonium)ethyl methanethiosulfonate (MTSET, 0.5 mM), a MTS reagent in which the amino group is replaced by a trimethylamine, failed to induce any detectable currents in ASIC3^E79C-expressing CHO cells (Figure 6C), suggesting that the amino group in AGM plays an essential role in activating ASIC3 channels, as has been shown in GMQ-ASIC3 interaction [10]. Together, these data strongly support that activation of ASIC3 by AGM relies on polar, steric, and electrostatic interactions with the nonproton ligand sensing domain in the channel [10], regardless whether the ligand is linear and flexible (i.e. AGM) or circular and rigid (i.e. GMQ).

Most ASIC-like acid-evoked currents in DRG neurons are mediated by heteromers of ASIC3, -2, and -1 [57]. To address ASIC subunit specificity, we recorded AGM or ARC responses in CHO cells expressing different combinations of ASIC subunits (Figure 7). Similar to GMQ [10], none of the homomeric channels ASIC1a, ASIC1b, or ASIC2a were activated by either AGM or ARC (Figure 7A, C). However, heteromeric ASIC3 plus ASIC1b channels responded to AGM and ARC (Figure 7B, D) in a manner similar to homomeric ASIC3 channels (Figure 7A, C). On the other hand, heteromeric combinations of ASIC3 plus ASIC1a, ASIC2a, or ASIC2b were insensitive to these ligands (Figure 7B, D). This subunit specificity together with the restricted distribution pattern of ASIC3 [9] suggests that an AGM-dependent regulatory pathway most likely occurs in peripheral tissues expressing homomeric ASIC3 and/or heteromeric ASIC3 + 1b channels [8, 36, 58]. Alternatively, AGM may act as a co-agonist sensitizing the apparently-unresponsive heteromeric ASIC3 channels (i.e., ASIC3 + 1a, 2a, or 2b) (Figure 7) to protons. In addition, according to a recent report [59], AGM may also act on the ASIC3 channels through coincident detection of multiple ligands (i.e., AGM, H⁺, and other factors such as ATP) by cross-activating an as yet unidentified ion channel. In addition, the emergence of new ASIC isoforms [60] adds an additional possibility underlying the ASIC3-dependent pain-behavior induced by AGM in vivo [10].

Finally, to gain insights into the pathophysiological relevance of ASIC3-dependent integration of AGM and multiple inflammatory signals, we performed in vivo pain-related behavioral tests [10] following the injection of AGM and hyperosmolarity, arachidonic acid (AA), or lactate into the right hindpaw of asic3^+/+ and asic3^-/- mice. We measured the total time the animals spent licking the injected paw during a 30-min period. As shown previously, control asic3 ^+/+ mice showed a significant increase in paw-licking time after AGM (10 mM) injection compared to saline-injected controls [10]. In consideration of the high AGM concentration used, we re-evaluated the paradigms by injecting 1 mM AGM (Figure 8). Similarly, asic3 ^+/+ mice showed a significant increase in paw-licking time after AGM (1 mM) injection, though less intense than that observed following 10 mM AGM injection [10]. As expected, the reaction of asic3^-/- mice to AGM was significantly reduced (Figure 8). When AGM (1 mM) was co-applied with hyperosmolarity (H-Osm, 600 mosmol kg^-1 with mannitol), AA (10 μM), or lactate (15 mM, keeping the pH neutral), the injection elicited more intense response in asic3 ^+/+ mice while failed to elicit the comparable response in asic3 ^-/- mice. Interestingly, asic3 ^+/+ mice showed a significant increase in paw-licking time following the treatment of H-Osm (600 mosmol kg^-1 with mannitol), AA (10 μM), or lactate (15 mM) alone compared to saline-injected controls. These behavioral responses were similarly reduced in asic3 ^-/- mice (Figure 8), supporting an essential role of ASIC3 in sensing multiple inflammatory signals [9], including AGM.

All constructs were expressed in CHO cells as described previously [10]. In brief, CHO cells were cultured at 37 °C in a humidified atmosphere of 5% CO₂ and 95% air. The cells were maintained in F12 medium (INVITROGEN) supplemented with 1 mM L-glutamine, 10% fetal bovine serum, 50 units/ml penicillin, and 50 μg/ml streptomycin. Transient transfection of CHO cells was carried out using Lipofectamine™2000 (INVITROGEN). Electrophysiological measurements were performed 24-48 h after transfection.

The ionic composition of the incubation solution (SS, see Figures 2E, F and 6B) was (mM): 150 NaCl, 5 KCl, 1 MgCl₂, 2 CaCl₂, 10 HEPES, and 10 glucose, aerated with 95% O₂/5% CO₂ to a final pH of 7.4. The standard external solution contained (mM): 150 NaCl, 5 KCl, 1 MgCl₂, 2 CaCl₂, and 10 glucose, buffered to various pH values with either 10 mM HEPES, pH 6.0-7.4, or 10 mM MES, pH < 6.0. For the Na⁺-free medium (Figure 6B), Na⁺ was substituted with equimolar N-Methyl-D-glucamine (NMDG). The patch pipette internal solution for whole-cell patch recording was (mM): 120 KCl, 30 NaCl, 1 MgCl₂, 0.5 CaCl₂, 5 EGTA, 2 Mg-ATP, and 10 HEPES. The internal solution was adjusted to pH 7.2 with Tris-base. The osmolarities of all these solutions were maintained at 300-325 mOsm (Advanced Instrument, Norwood, MA). Hyperosmotic (H-Osm) conditions were obtained by adding mannitol to the standard external solution (or saline) as indicating in the text.

Solutions with different composition were applied using a rapid application technique termed the "Y-tube" method throughout the experiments [10]. This system allows a complete exchange of external solution surrounding a cell within 20 ms.

The cDNA of rat ASIC3 was subcloned into the pEGFPC3 vector (Promega Corporation, Madison, WI, U.S.A.). Each mutant was generated with the QuikChange^® mutagenesis kit (Stratagene, La Jolla, CA) in accordance with the manufacturer's protocol using high-performance-liquid-chromatography-purified or PAGE-purified oligonucleotide primers (Sigma-Genosys, The Woodlands, TX). Individual mutations were verified by DNA sequence analysis, and the predicted amino acid sequences were determined by computer analysis.

The electrophysiological recordings were performed using the conventional whole-cell patch recording configuration under voltage clamp condition. Patch pipettes were pulled from glass capillaries with an outer diameter of 1.5 mm on a two-stage puller (PP-830, Narishige Co., Ltd., Tokyo, Japan). The resistance between the recording electrode filled with pipette solution and the reference electrode was 3-5 MΩ. Membrane currents were measured using a patch clamp amplifier (Axon 700A, Axon Instruments, Foster City, CA) and were sampled and analyzed using a Digidata 1320A interface and a computer running the Clampex and Clampfit software (version 8.0.1, Axon Instruments). In most experiments, 70-90% of the series resistance was compensated. Unless otherwise noted, the membrane potential was held at -60 mV throughout the experiment under voltage clamp conditions. All the experiments were carried out at room temperature (23 ± 2 °C).

Animals were acclimatized for 30 min before experiments. A total volume of 10 μl solution (in 0.9% NaCl) containing either saline (0.9% NaCl only), or AGM (1 mM), or hyperosmolarity (H-Osm, 600 mosmol kg^-1 with mannitol), or AA (10 μM), or lactate (15 mM), or AGM + H-Osm, or AGM + AA, or AGM + lactate was injected intraplantarly using a 30G needle and paw-licking behavior was quantified for 30 min [10].

Results were expressed as means ± S.E.M. Unless otherwise noted, statistical comparisons were made with the Student's t test.*, or ^&, or ^@, p < 0.05 or **p < 0.001 was considered significantly different. To test the synergic interaction between two factors (i.e., mild acidosis and AGM, or AGM and Ca²⁺ reduction) on ASIC3 currents, additional two-way ANOVA analyses were made (Figures 2 and 5, p < 0.05 was considered significant). Concentration-response relationships for pH-dependent activation of ASIC3 channels were obtained by measuring currents in response to acidic solutions with graded pH values. Each acidic solution was tested on at least three CHO cells and all results used to generate a concentration-response relationship were from the same group. The data were fit to the Hill equation: I/I _max =1/[1+(EC ₅₀/[Ligand]) ⁿ ], where I is the normalized current at a given pH, I _max is the maximum normalized current, EC ₅₀ is the concentration of proton yielding a current that is half of the maximum, and n is the Hill coefficient.