Multiple Modulation of Acid-Sensing Ion Channel 1a by the Alkaloid Daurisoline

Acid-sensing ion channels (ASICs) are proton-gated sodium-selective channels that are expressed in the peripheral and central nervous systems. ASIC1a is one of the most intensively studied isoforms due to its importance and wide representation in organisms, but it is still largely unexplored as a target for therapy. In this study, we demonstrated response of the ASIC1a to acidification in the presence of the daurisoline (DAU) ligand. DAU alone did not activate the channel, but in combination with protons, it produced the second peak component of the ASIC1a current. This second peak differs from the sustained component (which is induced by RF-amide peptides), as the second (DAU-induced) peak is completely desensitized, with the same kinetics as the main peak. The co-application of DAU and mambalgin-2 indicated that their binding sites do not overlap. Additionally, we found an asymmetry in the pH activation curve of the channel, which was well-described by a mathematical model based on the multiplied probabilities of protons binding with a pool of high-cooperative sites and a single proton binding with a non-cooperative site. In this model, DAU targeted the pool of high-cooperative sites and, when applied with protons, acted as an inhibitor of ASIC1a activation. Moreover, DAU’s occupation of the same binding site most probably reverses the channel from steady-state desensitization in the pH 6.9–7.3 range. DAU features disclose new opportunities in studies of ASIC structure and function.


Supplementary Materials
1 We simulated the kinetic of proton-activated currents at the ASIC1a channel 2 (taking into account the rate of change of the solution) using the method of 3 numerical solution of a system of kinetic equations using an algorithm similar to 4 that described [1]. The accuracy of this modeling method was verified by 5 comparing our calculations with analytical solutions for a three-state [2] and 6 four-state [3] models of receptor activation by the ligand. Comparison of 7 dose-response curves fitted to the maximum amplitude of the current generated by 8 our program with these analytical solutions showed complete coincidence of 9 numerical simulation predictions (the relative amplitude error was less than 10 −7 % 10 over the entire range of ligand concentrations (10 −9 -10 −3 M). Thus, our algorithm can 11 be successfully used to simulate the currents of ligand-activated receptors, 12 including rapidly activated, such as ASIC channels.

13
As Scheme S1. Model 1/2. Model 2 with introduced cooperativity factors (p and q < 1). 20 The kinetic constants kon = 2 × 10 9 М −1 s −1 and koff = 10 3 s −1 for protons binding to 21 ASIC1a were suggested early in literature by [6]. Other constant: The rate constant 22 of the channel opening α = 1000 s −1 ; the rate constant of the channel closing β = 1000 23 s −1 ; the rate constants of the channel transition to desensitization γ = 10 s −1 , and the 24 rate constants of the channel transition from desensitization ε = 0.02 s −1 , were chosen 25 on the basis of the speed and magnitude of the desensitization current decrease. 26 The coefficients "p" and "q", highlighted in red, reflect the cooperativity of the 27 interaction of protons with the channel and they were introduced into the Model 2 28 (see description below).

29
We investigated dose-dependence of the probability of the channel opening 30 for Model 1 with 3, 6, and 9 identical binding sites (M1-3, M1-6, M1-9). This pool of 31 sites number was chosen in view of homotrimeric ASIC1a organization. All three 32 models reproduced similar shape of the current generated by various proton 33

39
The pH-dependence of the channel activation was analyzed for all three 40 models ( Figure S1B). In all cases the dependences of the maximum amplitude of the 41 current versus proton concentration keep an asymmetrical shape of curve like the 42 experimental curve, and poorly fitted by the logistic equation. Otherwise a reliable 43 fitting of we fitted the dose-response with the following equation F2: 44 The dose-dependence activation of all models is perfectly fitted by this equation.

46
The Hill coefficients obtained by the fittings are shown in Table S1. 47 The value of the nH2 always remains less than one (as in the experimental 49 data). The value nH1, which determines the steepness of the left part of the 50 proton-dependence curve of activation ( Figure S1B) has a tendency to grow when a 51 number of proton sites in Model 1 is increased, but this tendency is not enough to 52 achieve higher nH1 values. Therefore, the Model 1 without modification failed to 53 obtain high nH1 values close to experimental (more than 6) by a simple increasing of 54 proton binding sites number. 55 In order to obtain higher nH1, we designed the Model 2 with two parametric 56 coefficients p and q. We assumed that the binding of protons is a cooperative 57 process, causing the conformational rearrangements in the channel that were 58 described [7] as a collapse of the pocket where protons bind. As a result, each 59 successive proton will have different binding/dissociation characteristics than the 60 previous one. Thus, we believed that the rate constants of both binding and 61 dissociation decrease with each subsequent process of interaction of the proton 62 with the channel and introduced the cooperativity coefficient p < 1 (for binding 63 constants) and coefficient q (for dissociation constants) that always less than p. 64 As a result, we managed to achieve an increase in the value of nH1 ( Figure  65 S2A, B). For the M2-6, nH1 does not exceed a value of 4.5 for any values of the "p" 66 and "q" coefficients ( Figure S2A). For the M2-9 model, the value of nH1, about 6.4, is 67 achieved only in a narrow range of q and p values. For further modeling, we take 68 the q p and values of the main maximum ( Figure S2B,C). Thus, we assume that the 69 number of proton-binding sites equal to nine is a necessary and sufficient, and best 70 calculation is produced for q coefficient of 0.65 and p coefficient of 0.75 ( Figure  71 S2D). It should be noted, that the M2-9 model with selected p and q values 0.75 and 79 0.65, respectively, give the satisfactory result for nH1 and nH2 but false pH501 and 80 pH502 values. Experimental pH501 is greater than pH502 (Figure 6a in main text) 81 while this model produce more high value for pH502 ( Figure S2D). To overcome this 82 inconsistency, we introduced into Model 2 an additional proton binding site with 83 and noncooperative less affine site (Model 3; see also the scheme and text in the 84 main results). 85

86
Scheme S2. Model 3 87 We introduced the following assumptions into this model: 88 1) The tenth site is low affine, k'on = 2 × 10 9 М −1 s −1 and k'off = 1.2 × 10 3 s −1 (the last 89 constant was taken based on the difference between pH501 and pH502 in the 90 experiment); 91 2) Proton binding in the main (highly cooperative) pool does not affect proton 92 binding in the low affinity region; 93 3) Proton binding in the low affinity site also does not affect proton binding in the 94 cooperative pool, i.e., p = r, q = s. 95 Under these assumptions, the entire dose-dependence is shifted to the right 96 in the region of acidic pH. Returning the curve to the left is possible by increasing 97 the cooperativity of the binding by increasing the coefficients of cooperativity p and 98 r. This allowed us to obtain a dose-dependence, which is fitted by equation F2 with 99 parameter values close to the experimental ones (see Table S2). 100 It should be noted that, although the parameters of the fitting of the model 103 agree well with the experimental data, the coefficient nH1 becomes smaller. This can 104 be compensated for by increasing the number of highly operational sites, or by 105 fixing the second coefficient when fitted by equation F2. Since the experimental data 106 showed the second coefficient very close to 1, we adjusted the data of the M3-10 107 model with the coefficient nH2 fixed to the value 1. These fittings are shown in the 108 main text and in Figure 6. 109