Computational Investigation of Mechanisms for pH Modulation of Human Chloride Channels

Abstract

Many transmembrane proteins are modulated by intracellular or extracellular pH. Investigation of pH dependence generally proceeds by mutagenesis of a wide set of amino acids, guided by properties such as amino-acid conservation and structure. Prediction of pKas can streamline this process, allowing rapid and effective identification of amino acids of interest with respect to pH dependence. Commencing with the calcium-activated chloride channel bestrophin 1, the carboxylate ligand structure around calcium sites relaxes in the absence of calcium, consistent with a measured lack of pH dependence. By contrast, less relaxation in the absence of calcium in TMEM16A, and maintenance of elevated carboxylate sidechain pKas, is suggested to give rise to pH-dependent chloride channel activity. This hypothesis, modulation of calcium/proton coupling and pH-dependent activity through the extent of structural relaxation, is shown to apply to the well-characterised cytosolic proteins calmodulin (pH-independent) and calbindin D9k (pH-dependent). Further application of destabilised, ionisable charge sites, or electrostatic frustration, is made to other human chloride channels (that are not calcium-activated), ClC-2, GABA_A, and GlyR. Experimentally determined sites of pH modulation are readily identified. Structure-based tools for pKa prediction are freely available, allowing users to focus on mutagenesis studies, construct hypothetical proton pathways, and derive hypotheses such as the model for control of pH-dependent calcium activation through structural flexibility. Predicting altered pH dependence for mutations in ion channel disorders can support experimentation and, ultimately, clinical intervention.

Keywords: bestrophin; calcium binding; chloride channels; pH dependence; protein electrostatics.

Figure 1

Electrostatic environment of the cBest1 calcium clasp. (A) Calcium clasp location in cBest1. (B) Calcium ion location (red sphere) and colour-coded representation for ΔpKas of pH-titratable sidechains adjacent to the calcium site in the holo structure (6n27). Although calcium is displayed, pKa calculations are made without it, to enable direct comparison with the apo structure. Red sidechains denote destabilisation of the ionised form and electrostatic frustration, and blue are stabilised. (C) The calcium-binding loop is disordered in the apo structure (6n26), with the calcium ion location copied over from the holo form for reference.

Figure 2

Electrostatic environment of the bBest2 calcium clasp. (A) Clasp location in bBest2. (B) pH-titratable sidechain colour coding, and ΔpKa calculations as for Figure 1, in the holo structure (6vx7). (C) Structural relaxation of the calcium-binding region and reduction of carboxylate sidechain electrostatic frustration in the apo form (6vx8).

Figure 3

Electrostatic environment of calcium binding sites in calmodulin and calbindin D_9k. Two calcium (EF-hand) sites are shown in each plot, with ionisable group colour-coding from the Protein-Sol pKa tool (see Figure 1). Sidechain residue labels are only given for one of the two sites, in each case. (A) Holo calmodulin 1up5 [30]. (B) Apo calmodulin 1cfd [26]. (C) Holo calbindin 4icb [31]. (D) Apo calbindin 1clb [32].

Figure 4

Sites of pH sensing in the ClC family. (A) Cartoon tube representation of human ClC-1 dimer structure (pdb 6qvu [46]), with the location of H555 indicated in one monomer. (B) An expanded view of the packing environment around the H555 pH sensor, with groups labelled, and a destabilising pKa from Protein-Sol denoted (red). (C) Protein-Sol output is visualised (PyMol) for a monomer AlphaFold model of human ClC-2. Groups predicted to mediate pH-dependent effects around neutral pH are highlighted and labelled.

Figure 5

Charge cluster around the GABA binding site of the GABA_A receptor. (A) Side view (along the membrane) of GABA_A receptor with bound Fab fragments labelled, structure pdb 6d6u, [55]. Colouring is by chain with residues linked to pH sensing shown as yellow spheres, GABA as green spheres and flumazenil as red spheres. (B) Top view (into the membrane) with subunits labelled. (C) GABA binding site (green spheres) with residues linked to pH sensing shown as yellow spheres and residues predicted to be electrostatically frustrated residues as red sticks.

Figure 6

Electrostatic frustration in GlyR (5cfb) [63] identifies functional sites of pH sensing and electrostatic repulsion. Protein-Sol calculated and colour-coded ΔpKas (red destabilising, blue stabilising) are illustrated for a single protomer of the pentamer with spheres on ionisable group Cα atoms. Two neighbouring monomers are also displayed (cartoon representation), leaving a ‘cut-away’ view of the channel, which is displayed along the membrane (left-hand side) and with a tilted view (from the intracellular side) to indicate the channel pore. Groups predicted to be electrostatically frustrated are labelled and indicated with arrows on the left and righthand plots.