Elucidating the role of the intracellular pH sensing mechanism of TASK-2 K2P channel

Two-pore domain potassium (K2P) channels are responsible for maintaining the background conductance essential to the resting membrane potential. K2P channels assemble as dimers containing two pore-forming domains and four transmembrane segments per subunits. Two fenestrations connect the lipid membrane with the central conduction cavity, which can be open or closed depending of the movements of helix TM4. TALK subfamily of K2P channels is activated by alkaline extracellular pH and is formed by 3 members: TALK-1, TALK-2 and TASK-2. TASK-2 is also gated by intracellular pH (pHi), being closed by intracellular acidification and activated by increasing pHi. The neutralization of lysine positioned at the end of TM4 helix, and probably within the fenestrations, by a mutation to K245A abolishes pHi-gating . The molecular mechanism by which pHi-sensing K245 exerts its gating role is unknown. A possible mechanism suggest that K245 protonated is able to open the fenestrations and therefore close the channel. Through computational studies, we modeled the 3D structure of TASK-2 channel in both fenestration states, these models were used as a starting point to perform molecular dynamics simulations. The trajectories analysis reveals a good agreement between the pK1/2 of K245 obtained experimentally and the pKa predicted MOL2NET, 2018, 4, http://sciforum.net/conference/mol2net-04 2 when the fenestrations are closed. Besides, we proved that Norfluoxetine compound is a potent blocker of TASK-2 channels and its putative binding site is within the fenestrations (data not shown).


Introduction
Two-pore domain potassium (K2P) channels take part in stabilize the negative resting membrane potential in excitable cells. To date, 15 mammalian genes codifying K2P channels have been identified, which are classified into 6 subfamilies 1 : TWIK, THIK, TRAAK, TRESK, TASK and TALK. Each K2P channel subunit contains two pore forming domains and four transmembrane segments (TM1-TM4) and they assembly functionally as dimers. Two unusual openings called fenestrations were discovered in crystallographic structures, which connect the lipid membrane with the central conduction cavity of K2P channels. The elucidation of TRAAK channel crystallographic structures by Brohawn 2  an atomistic explanation about the K245 pHi sensor due to the proximity of K245 to these hydrophobic fenestrations: "the protonated state of K245 (K245 + ) within of the fenestration promotes their opening and therefore the closure of the TASK-2 channel". Through the Niemeyer's hypothesis is suggested the presence of an inner gate in TASK-2, which could be related with the state of the fenestrations. However, in TASK-2 channel, the inner gate has not been investigated directly, mainly due to a lack of high-affinity TASK-2 blockers that binds within the fenestrations.

Materials and Methods
Homology Modelling: • The complete sequence of human TASK-2 was downloaded from Uniprot (ID: O95279).

Molecular Dynamics simulation (MDs):
• The TASK-2 models were prepared to perform MDs with the Schrödinger 9 program. Thus, for each model two system were built: 1) with the intracellular pH sensor K245 protonated (pH = 7.5) and 2) neutral. The neutral state of K245 was predicted computationally using PropKa3.0 program 10 . • The TASK-2 systems were embedded into a pre-equilibrated POPC membrane and solvated in a periodic box of SPC water molecules, then the systems were neutralized by adding 150 mM of NaCl. • The systems were subjected to an energy minimization and 100 ns of MDs employing OPLS 11 as force-field and thus correct the errors inherent in the modeling step. Only secondary structure restraints were applied of 0.2 kcal mol -1 Å -2 .

Results and Discussion
The trajectory analysis reveals that all models are thermodynamically stables under 3 Å of root mean square deviation (RMSD). Furthermore, TASK-2 based on TRAAK (3um7) with K245 + (black line), TRAAK (4wff) with K245 + (blue line) and TREK-2 (4xdk) with K245 + (red line), are more stables than rest of the models based on TRAAK (3um7, gray line), TRAAK (4wff, cian line), and TREK-1 (5vkp, magenta line) with K245 neutral (K245 0 ). The most variable RMSD is for TASK-2 based in TREK-1 (5vkp) because is the only model including the C-terminal region of the channel, being the loops of this region the main contributors to the RMSD fluctuation.
The pKa prediction of K245 in TASK-2 calculated with PropKa3.0 shown that the nearest values to the experimental pK1/2 (~ 8.0) are obtained when the fenestrations are closed, and these are: TASK-2 based in TREK-1 (5vkp) in both monomers, in TREK-2 (4wff) only monomer A (with the TM4 helix in up-state and therefore the fenestration closed) and TREK-2 (4bw5) in both monomers. All the predicted pKa values were calculated as an average over 200 ns of MD simulations evaluating 1 frame per ns (n=200).

Conclusions
Using comparative modelling techniques and different templates, it was possible to obtain the relative position of the intracellular pH sensor of TASK-2: K245, regarding to both conformational states of the fenestrations (open & close). The computational prediction of the pKa of K245 over a MD trajectory (n=200 structures) of all comparative models suggest that the pK1/2 of K245 obtained experimentally was made over the channel with the fenestrations closed, in agreement with the Brohawn 3 hypothesis.