**3. Discussion**

In the present work we demonstrated that many hydrophobic monoamines and their guanidine analogs affected ASIC3 in submillimolar concentrations. Two of them, IEM-2163 and IEM-2195, were studied in detail, whereby we found that their effects are best explained by the existence of two distinct mechanisms. The first mechanism is the acidic shift of activation that results in fast pH-dependent peak inhibition. The second one is deceleration of the ASIC3 desensitization, which raises the equilibrium level of the sustained current, thus effectively increasing its amplitude. The total drug effect on the sustained current depended on the ratio of these two independent types of actions.

We found that in a large series of drugs there was no correlation between these two types of action. Effect on activation was found to be pH-dependent, whereas modulation of desensitization was similar at pH 6.85 and pH 6.0. Elucidation of these types of action in turn required different application protocols. Concentration dependencies were also apparently separated, with effect on activation developing at lower concentrations than the effect on desensitization. There was also a drastic difference in kinetics, as the effect on activation was much faster than on desensitization. Taken together, these data suggest that two distinct types of action are mediated by drugs binding to different sites.

Drug effects on ASIC3 have been the subject matter of numerous studies. For instance, a detailed examination of GMQ and a representative series of its derivatives [44] allows for comparison with our data. In particular, compounds containing two aromatic rings and a guanidine group used at a concentration of 1 mM induced an acidic shift in the activation curve of ASIC3, similarly to IEM-2163, IEM-2195, and some other compounds in our work (Figure 2). On the other hand, GMQ and a few other derivatives induced an alkaline shift of activation, while we found no such effect for our compounds. Amiloride, a known ASIC blocker, also causes an alkaline shift of activation in high concentrations (0.5–1 mM). [24]. A similar but much weaker effect was induced by agmatine [26].

Notably, to detect a shift of activation it is necessarily to study the ligands' effects with both weak and strong acidifications, and such data are not available for a number of other compounds.

Analysis of drugs' action on sustained current is more complex, as it can be mediated by the effects on both channel activation and desensitization. Additionally, in some experimental setups ASIC3 does not mediate such currents in control, complicating quantitative estimations of effects. Various drugs, including GMQ [23], agmatine [26], and amiloride [22], induce or potentiate sustained ASIC3-mediated currents evoked by modest acidifications. Serotonin [27] and FMRFamide [25] potentiate the sustained current, while simultaneously slowing down desensitization kinetics under conditions of strong acidification. Note that these two compounds did not affect the peak component. In the present work we showed that drugs reduced the speed of response decay and increased final equilibrium level of sustained current amplitude under conditions of modest acidifications. We also experimentally separated this effect from their influence on activation, allowing us to detect such an effect for IEM-2163 despite its total inhibitory action. We are not aware of the proven examples of the compounds inhibiting the sustained current via modulation of desensitization.

In this regard it is interesting to compare drugs' effect on ASIC1a and ASIC3. In our previous paper [32] we demonstrated that many monoamines and their guanidine analogs affect the steady-state desensitization of ASIC1a by shifting its pH dependence to more acidic values, although this effect does not lead to the appearance of sustained current. The opposite effect, alkaline shift of the steady-state desensitization, was not revealed for small molecules but only for psalmotoxin [45]. Thus, if we assume that a similar process underlies desensitization in both ASIC1a and ASIC3, there is an apparent commonality in the direction of drug action, although it manifests differently, according to the channel type. In contrast, the drug action on activation properties is notably diverse. For instance, IEM-2044 and amitriptyline have opposite effects on different channels, inducing an alkaline shift of activation on ASIC1a [32,34] and an acidic one on ASIC3, while 9AA shifts the activation to more acidic values in both cases [32]. Histamine only enhanced the activation of ASIC1a and was inactive against ASIC3 [33]. Thus, we do not see a correlation for action on activation of ASIC1a and ASIC3. Similarly, in [44], GMQ and its derivatives also demonstrated varying effects on ASIC1a and ASIC3, with some compounds acting differently on different channels and others having the same effect regardless of the target.

The problem of the binding site(s) of ASIC ligands in the extracellular domain is intensely debated. According to recent structural data, channel "activation involves 'closure' of the thumb domain into the acidic pocket, expansion of the lower palm domain and an iris-like opening of the channel gate. The linkers between the upper and lower palm domains serve as a molecular 'clutch', and undergo a simple rearrangement to permit rapid desensitization" [46]. Another study [47] suggests that the protonable residues in the acidic pocket affect ASIC pH dependence, but in the palm domain they are responsible for the regulation of desensitization kinetics as well as prevention of the sustained currents. Thus, different regions participate in complex allosteric interactions which contribute to activation and desensitization, which in turn significantly complicates estimation of ligands' binding site(s). In addition, particular mutations can unequally affect different modes of ligands' action. One such example is Glu-79 in the palm domain of ASIC3 [48]. While it has been shown to be a crucial element for direct opening of the channel by GMQ, its mutation did not elicit any changes in GMQ's effects on activation, but instead altered GMQ's influence on the channel inactivation. The effect of mutations on ligands' binding and action can also be either direct or allosteric. These data, together

with the complex structure–activity relationships revealed in the present and other studies, raise the possibility that low-weight drugs can bind to more than one site in the extracellular domain of ASICs. For instance, binding to the acidic pocket could control effects on activation, whereas binding to the palm domain could be responsible for desensitization effects.

In our work we have focused our attention primarily on the low-to-moderate (pH 6.85–6.0) acidification range. While the more powerful acidification (pH < 5.0) that is frequently used in other studies can indeed occur [49] in vivo, it typically accompanies severe conditions such as tumors and open fractures. However, physiological processes and less-drastic pathologies usually stay in the less-acidic pH range [50–53]. Additionally, the research of Salinas et al. [54] indicates that ASIC3 activations by different levels of acidification are facilitated by distinct mechanisms. If one were to assume that those mechanisms in turn mediate specific physiological responses, then our work shows potential for the development of state-dependent drugs, which would affect only the specific response, without influencing other channel functions.
