Ionic Coulomb blockade and anomalous mole fraction effect in NaChBac bacterial ion channels
I.Kh. Kaufman, O.A. Fedorenko, D.G. Luchinsky, W.A.T. Gibby, S.K. Roberts. P.V.E. McClintock, R.S. Eisenberg
NNoname manuscript No. (will be inserted by the editor)
Ionic Coulomb blockade and anomalous mole fractioneffect in NaChBac bacterial ion channels
I.Kh. Kaufman · O.A. Fedorenko · D.G. Luchinsky · W.A.T. Gibby · S.K. Roberts · P.V.E. McClintock · R.S. Eisenberg the date of receipt and acceptance should be inserted later
Abstract
We report an experimental study of the influences of the fixed chargeand bulk ionic concentrations on the conduction of biological ion channels, and weconsider the results within the framework of the ionic Coulomb blockade modelof permeation and selectivity. Voltage clamp recordings were used to investigatethe Na + /Ca anomalous mole fraction effect (AMFE) exhibited by the bacterialsodium channel NaChBac and its mutants. Site-directed mutagenesis was usedto study the effect of either increasing or decreasing the fixed charge in theirselectivity filters for comparison with the predictions of the Coulomb blockademodel. The model was found to describe well some aspects of the experimen-tal (divalent blockade and AMFE) and simulated (discrete multi-ion conductionand occupancy band) phenomena, including a concentration-dependent shift ofthe Coulomb staircase. These results substantially extend the understanding ofion channel selectivity and may also be applicable to biomimetic nanopores withcharged walls. Keywords
Ionic Coulomb blockade, NaChBac, bacterial channel, electrostaticmodel, BD simulations, site-directed mutagenesis, patch-clamp technique
Biological ion channels are natural nanopores providing for the fast and highly se-lective permeation of physiologically important ions (e.g. Na + , K + and Ca ) I.Kh. Kaufman, W.A.T. Gibby, P.V.E. McClintockDepartment of Physics, Lancaster University, Lancaster LA1 4YB, UKE-mail: [email protected]. Fedorenko, S.K. RobertsDivision of Biology and Life Sciences, Lancaster University, Lancaster UKD.G. LuchinskyDepartment of Physics, Lancaster University, Lancaster LA1 4YB, UK and SGT, Inc., Green-belt, MD, 20770, USAR.S. EisenbergDepartment of Molecular Biophysics, Rush University, Chicago, IL, US a r X i v : . [ q - b i o . S C ] D ec I.Kh. Kaufman et al. through cellular membranes [1–3]. Despite their fundamental importance, andnotwithstanding enormous efforts by numerous scientists, the physical origins oftheir selectivity still remain unclear. It is known, however, that the conductionand selectivity properties of cation channels are defined by the ions’ movementsand interactions inside a short, narrow selectivity filter (SF) lined by negativelycharged amino acid residues that provide a net fixed charge Q f [1, 2].NaChBac bacterial sodium channels [4–6] are frequently thought of, and usedas, simplified experimental/simulation models of mammalian calcium and sodiumchannels. X-ray investigation and molecular dynamics simulations have shown thatthese tetrameric channels possess strong binding sites with 4-glutamate { EEEE } loci at the SF [7]. Bacterial channels have been used in site-directed mutagenesis(SDM) /patch clamp studies of conductivity and selectivity [6, 8].Conduction and selectivity in calcium/sodium ion channels have recently beendescribed [9–11] in terms of ionic Coulomb blockade (ICB) [12, 13], a fundamentalelectrostatic phenomenon based on charge discreteness, an electrostatic exclusionprinciple, and single-file stochastic ion motion through the channel. Earlier, VonKitzing had revealed the staircase-like shape of the occupancy vs site affinity forthe charged ion channel [14] (following discussions and suggestions in [15]), andcomparable low-barrier ion-exchange transitions had been discovered analytically[16]. A Fermi distribution of spherical ions was used as the foundation of a PoissonFermi theory of correlated ions in channels [17, 18].ICB has recently been observed in sub-nm nanopores [13]. It appears to beclosely similar to its electronic counterpart in quantum dots [19]. As we havedemonstrated earlier [11], strong ICB appears for Ca ions in model biologicalchannels and manifests itself as an oscillation of the conductance as a function of Q f , divalent blockade, and AMFE.Here we present an enhanced model of divalent blockade and AMFE able toencompass concentration-related shifts in the ICB conduction bands and the shapeof the divalent blockade decay. We compare model predictions with the literature,with our own earlier simulated data [10, 11], and with new experimental resultsfrom a patch clamp study of conductivity and selectivity in the NaChBac bacterialchannel and its mutants that has enabled Q f to be changed. ICB-based Q f vs. log[ Ca ] phase diagrams are introduced to explain visually the differences betweenthe AMFE behaviours observed for different mutants, where [ Ca ] is the Ca ionconcentration.In what follows ε is the permittivity of free space, e is the proton charge, z isthe ionic valence, T the temperature and k B is Boltzmann’s constant. Figure 1 summarises the generic, self-consistent, electrostatic model of the selec-tivity filter of a calcium/sodium channel introduced earlier [10]. It consists of anegatively-charged, axisymmetric, water-filled, cylindrical pore through the pro- tein hub in the cellular membrane; and, we suppose it to be of radius R = 0 . L = 1 . + /Ca channels.There is a centrally-placed, uniformly-charged, rigid ring of negative charge0 ≤ | Q f /e | ≤ R Q = R to represent the charged onic Coulomb blockade and AMFE in NaChBac bacterial ion channels 3 Fig. 1
Generic electrostatic model of Calcium/Sodium ion channel [10]. The model describesthe selectivity filter of ion channel as an axisymmetric, water-filled pore of radius R = 0 . L = 1 . Q f is embedded in the wall to represent the charged residues of realCa /Na + channels. We take both the water and the protein to be homogeneous continuadescribable by relative permittivities ε w = 80 and ε p = 2, respectively, together with animplicit model of ion hydration whose validity is discussed elsewhere. The moving monovalentNa + and divalent Ca ions are assumed to obey self-consistently both Poisson’s electrostaticequation and the Langevin equation of motion. protein residues of real Ca /Na + channels. The left-hand bath, modeling theextracellular space, contains non-zero concentrations of Ca and/or Na + ions.For the Brownian dynamics simulations, we used a computational domain lengthof L d = 10 nm and radius R d = 10 nm, a grid size of h = 0 .
05 nm, and a potentialdifference in the range 0 −
25 mV (corresponding to the depolarized membranestate) was applied between the left and right domain boundaries. We take boththe water and the protein to be homogeneous continua describable by relativepermittivities ε w = 80 and ε p = 2, respectively, together with an implicit modelof ion hydration whose validity is discussed elsewhere [10].Of course, our reduced model represents a significant simplification of the ac-tual electrostatics and dynamics of ions and water molecules within the narrowselectivity filter due to, for example: the application of continuum electrostatics;the use of the implicit solvent model; and the assumption of 1D (i.e. single-file)movement of ions inside the selectivity filter. The validity and range of applicabilityof this kind of model have been discussed in detail elsewhere [10, 11, 21].This simplified self-consistent model was used as the basis for development ofthe ICB model of permeation and selectivity [10, 11], which led to some predictionsthat we now test experimentally in two complementary ways: through site-directedmutagenesis and patch-clamp studies of the bacterial sodium NaChBac channel; and numerically through Brownian dynamics simulations.The main aim was to test the ICB model’s predictions of the dependenceof the conductivity type, and of the divalent blockade/AMFE properties, on thefixed charge Q f at the SF. Site-directed mutagenesis and patch clamp measure-ments were used to investigate changes in the ion transport properties of NaChBac I.Kh. Kaufman et al.
Table 1
Main properties of the wild type (LESWAS) and mutant (LASWAS and LEDWAS)NaChBac bacterial channels generated and used for the present patch-clamp study. Here Q f stands for the nominal fixed charge at the selectivity filter and IC is the the [ Ca ] thresh-old value providing 50% blockade of the Na + current. Qualitative properties (selectivity andAMFE) are marked as “+” where present and “ − ” where absent. Channel SF amino acidsequence Nominal Q f /e Ca/Na selec-tivity Ca/Na AMFENaChBacwild-type L ES WAS − > Ca) − Zero-chargemutant L A SWAS 0 − −
Added-chargemutant LE D WAS − > Na) + (IC =5 µ M)mutants caused by alterations in the amino acid residues forming the SF, i.e. alter-ations in Q f . Increasing the value of Q f was expected to lead to stronger divalentblockade following the Langmuir isotherm and to a resonant variation of the diva-lent current with Q f [11]. The SF of NaChBac is formed by 4 trans-membrane segments each containing the six-amino-acid sequence LESWAS (leucine/glutamic-acid/serine/tryptophan/alanine/serine, respectively, corresponding to residues 190 - 195). This structureprovides the highly-conserved { EEEE } locus with a nominal Q f = − e [4]. Assummarised in Table 1, site-directed mutagenesis was used to generate two mutantchannels in which the SF either has “deleted charge” Q f = 0 (L A SWAS, in whichthe negatively charged glutamate E191 is replaced with electrically neutral alanine)or has “added charge” Q f = − e (LE D WAS, in which the electrically neutralserine S192 is replaced by negatively charged aspartate D [4].Details of the methods used for preparation of the mutants, and for the electro-physiology measurements, are presented in Appendix 1.
Coulomb blockade (whether ionic or electronic) arises in low-capacitance, discrete-state systems for which the ground state { n G } with n G ions in the channel isseparated from neighbouring { n G ± } states by a deep Coulomb gap U s (cid:29) k B T ,so that we can define the strength of the ICB as S ICB = U s / ( k B T ). The ICB phe-nomenon manifests itself as multi-ion oscillations (alternating conduction bandsand stop bands) in the Ca conductance and channel occupancy [9, 11].Figure 2 summarises the results of our earlier [9] Brownian dynamics simula-tion of Ca conduction, which was found to occur in multi-ion bands: A shows strong oscillations of conductance; and B the corresponding occupancy P , whichwas found to take the form of a Coulomb staircase where the steps in P occurin between resonant conduction points M n and current blockade points Z n , aspredicted by the ICB-based linear response model. Closer inspection of figure 2Bshows that the Coulomb staircase exhibits small concentration-related shifts [11]. onic Coulomb blockade and AMFE in NaChBac bacterial ion channels 5 J / J m a x P f /e| [ C a ], mM Fig. 2
Multi-ion conduction/occupancy bands in the model calcium channel, showing occu-pancy shifts with ionic concentration. A. Multi-ion calcium conduction bands M n as establishedby Brownian dynamics simulations. B. The corresponding Coulomb staircase of occupancy P c for different values of the extracellular calcium concentration [ Ca ], as marked, consists of stepsin occupancy as [ Ca ] changes. The neutralized states Z n providing blockade are interleavedwith resonant states M n . (Plots A, B are taken from [10]) C. Coulomb blockade-based phasediagram. The positions of the { n } → { n +1 } transitions (from equation (8)) are shown as blackdashed lines. The horizontal coloured lines are guides to the eye, indicating the three concen-trations used in the simulations. The diagram is consistent with the logarithmic [ Ca ]-relatedshift of steps in the Coulomb staircase shown in B. We now present a simple model to account for this shift, leading to the phasediagram shown in Figure 2C.We define the positions of the resonant conduction M n points (where barrier-less conduction can occur because G n = G n +1 , where G n is the Gibbs free energywhen there are n ions in the SF) taking account of concentrations P b and P c .We assume that the blockade is strong and we approximate U n by the dielectricself-energy of the excess charge Q n [11]: U n = Q n C ; Q n = zen + Q f ; C = 4 πε ε w R L (1)Here, C stands for the geometry-dependent self-capacitance of the channel .In equilibrium, the chemical potentials in the bulk µ b and in the channels µ c are equal[22, 23]: µ b = µ b, + k B T ln P b (Chemical potential of ions in the bath) (2) µ c = µ c, + ∆µ c,ex (Chemical potential of ions in the SF) (3) µ = µ b = µ c (Equibrium condition) (4)where the standard potentials µ b, and µ c, are assumed to be zero (although otherchoices are possible, [11]), P b and P c = (cid:104) n (cid:105) stand for equivalent bulk, and the SF I.Kh. Kaufman et al. occupancy is related to the SF volume V SF = πR L , i.e. P b = n b V SF , where n b isbulk number density of the species of interest.The excess chemical potential in the SF, ∆µ c,ex , is defined here as the excessGibbs free energy ∆G n = ∆U n − T ∆S n in the SF due to the single-ion { n } → { n +1 } transition, from (1). The SF entropy-related term T ∆S n is model-dependent,varying between the extremes for correlated motion and for an ideal gas [22, 24].We use the “single-vacancy” model of the motion [11, 25] for which the followingresult can be derived [24]: ∆G n = ∆U n + k B T ln( n + 1) (5)Hence, the equilibrium ( µ b = µ c ) occupancy around the transition point M n represents a thermally-rounded staircase (see Figure 2B) described by a Fermi-Dirac distribution [11, 16]: P ∗ c = P c − n = (cid:20) (cid:18) ( ∆G n − µ b ) k B T (cid:19)(cid:21) − = (cid:20) P b exp (cid:18) ∆G n k B T (cid:19)(cid:21) − (6)It corresponds to the Coulomb staircase (Figure 2B), well-known in Coulombblockade theory [19] which appears when varying either Q f or log( P b ).The resonant value M n of Q f for the { n } → { n + 1 } transition is defined as: M n = − ze ( n + 1 / − δM n ; (Nominal transition point) (7) δM n = ze C k B Tz e [ln( P b ) − ln( n + 1)] (Concentration-related shift) (8)Next we introduce the notion of “phase diagrams” and use them to describe theconcentration-related shifts seen in our earlier Brownian dynamics simulations(Figure 2B) and the divalent blockade/AMFE in mutation experiments on thebacterial NaChBac channel that we report below.The phase diagrams (Figure 2C, Figure 4C) represent the evolution of thechannel state on a 2-D plot with occupation log( P b ) (or equivalently log([ Ca ])concentration) on the ordinate axis and Q f /e on the abscissa or vice versa . Thephase transition lines (black, dashed) separate the states of the SF having differ-ent integer occupancy number { n } . Different sections through the diagram reflectdifferent experiments/simulations in the sense that we can choose to vary eitherthe concentration (divalent blockade/AMFE experiments) or Q f (patch clampexperiments on mutants).Let start from the concentration-related shift of the Coulomb staircase. Figure2C shows the switching lines and AMFE trajectory (projection of system evolu-tion) in the Ca ionic occupancy phase diagram [16] for the calcium/sodium chan-nel while Figure 2B shows the small concentration-related shifts of the Coulombstaircase for occupancy P found in the Brownian dynamics simulations [9]. Equa-tions (8) and the phase diagram provide a simple and transparent explanation ofthe simulation results. The origin of the shift lies in the logarithmic concentrationdependence of δM n in (8). Similar shifts were seen in earlier simulations [14]. Note that the BD simulations seem not to show any significant shift for the M pointswith increasing log([ Ca ]), an unexpected result that requires further investigation.Our electrophysiological measurements on NaChBac wild type LESWAS chan-nels and their mutants show what happens in reality. The net currents through amacroscopic number of identical biological channels can be resolved and displayed onic Coulomb blockade and AMFE in NaChBac bacterial ion channels 7 Fig. 3
Permeability to Na + and Ca of the wild type NaChBac (LESWAS) ion channel andits mutants (LASWAS and LEDWAS). A: Representative whole cell current vs time recordsobtained for channels in bath solution containing 140 mM Na + (left) or 100 mM Ca (right).B: Current-voltage I-V relationships ( ± SEM are shown as bars, n = 6 − + solution (red squares) and Ca solution (red triangles) and LEDWAS in Na + so-lution (green squares) and Ca solution (green triangles) normalized to the maximal peakNa + current from the same cell. C: Permeability Na + /Ca ratios determined using reversalpotentials, as described in [26], indicate that LEDWAS is a Ca selective channel. on a biologically relevant time scale. Figure 3A shows the original current tracesusing bath solutions containing either Na + or Ca as the charge carrying cation(see Appendix 1 for details of methods used). Zero-charge mutants ( Q f = 0) did not show any measurable current in either of the solutions, corresponding wellwith the Coulomb-blockaded state expected/measured for an uncharged channelor nanopore [11, 13].Wild type (LESWAS) channels ( Q f = − e ) exhibited high Na + conductance inagreement with earlier observations [4, 5] and with the ICB model which predicts I.Kh. Kaufman et al. J / J m a x −6 −4 −2 0 220406080 log ([Ca]), free, mM E r e v , m V −6 −4 −2 0 20123456789 log ([Ca]), free, mM Q f / e A CB n = 4n = 3n = 2n = 1n = 0IC IC Ca Na + Fig. 4
Divalent blockade and anomalous mole fraction effect (AMFE) in wild-type NaChBac(LESWAS, shown as red squares) and LEDWAS (green triangles) mutant channels. The bathsolution containing Na + and Ca cations was adjusted by replacement of Na + with equimolarCa ; the free Ca concentrations [ Ca ] are shown on the abscissa; and error bars representthe standard error in the mean (SEM). A: Averaged normalized peak currents of LESWAS andLEDWAS channels. The Langmuir isotherm (11) fitted to the LEDWAS data for [ Ca ] < E rev ) obtained from the samerecordings as in A indicate that LEDWAS stopped conducting Na + if [Ca ] ≥ ] (green) and [Na + ] (red) are fixed to100mM. C: Cartoon phase diagram Q f vs log([Ca]), where the switching lines predicted byEquation (8) are dashed-black. relatively Q f -independent Na + conduction due to the small valence z = 1 of Na + ions.LEDWAS channels with nominal Q f = − e were found to conduct both Na + and Ca (figure 3A). These results for LESWAS and LEDWAS are consistentwith previous reports [4, 6, 27]. They are also in agreement with the Coulombblockade model, which predicted conduction bands for divalent cations in thesemutants.To study the selectivity between Na + and Ca in more detail and to inves-tigate divalent blockade and AMFE, we performed experiments using bath solu-tions containing mixtures of Na + and Ca , at different concentrations. Ca wasadded to a bath solution (containing 140 mM Na + ) to achieve free Ca con-centrations from 10 nM up to 1 mM, which were calculated by Webmaxc ( http: //web.stanford.edu/~cpatton/webmaxcS.htm ) and achieved by adding HEDTA (for concentrations from 1 mM to 10 µ M) or EGTA (for concentrations ≤ µ M).Figure 4A also shows that the current through the “added charge” mutantchannel, LEDWAS, was highly sensitive to the presence of Ca , and fell rapidlywith increasing [ Ca ], i.e. it exhibited strong Ca blockade of its Na + currents.This first part of the AMFE phenomenon is well-known for calcium channels as onic Coulomb blockade and AMFE in NaChBac bacterial ion channels 9 divalent blockade [5, 28]. The blockade shape is frequently fitted empirically with aLangmuir isotherm, similarly to the cases of blockade by dedicated channel blockerdrugs [5].A complete description of divalent blockade and AMFE should account forstatistical and kinetic features of the multi-species solution inside the SF [17, 24].We use a simplified description based on the assumptions:[ Ca ] (cid:28) [ N a ]; τ Ca (cid:29) τ Na ; P c ([ Ca ]) + P c ([ N a ]) ≤ τ Ca and τ Na stand for the respective ionic binding times. Under these as-sumptions, the SF can be in two exclusive states: “open” ( P c ([ Ca ]) = 0 , J [ Na ] ([ Ca ]) = J [ Na ] (0)); and “closed”, blocked by Ca ions, ( P c ([ Ca ]) = 1, J [ Na ] ([ Ca ]) = 0),and the states are shared in time. Hence due to the ergodic hypothesis the blockadeof the Na + current reflects Ca occupancy: J Na ([ Ca ]) = J Na (0)(1 − P c ( Ca )) (10)The ICB model [11] predicts that blockade by Ca (or any other strong blocker)can be described by the Langmuir isotherm: X ([ Ca ]) = ln (cid:18) J ([ Ca ]) J (0) − J ([ Ca ]) (cid:19) = ln (cid:18) − P ∗ c P ∗ c (cid:19) = ln( IC ) − ln([ Ca ]) (11)where the monovalent partial current J ([ Ca ]) as a function of the bulk concen-tration [ Ca ]) is described by a Fermi-Dirac function (6) that is equivalent to theLangmuir isotherm (11) and IC is Ca concentration when J ( IC ) = 0 . J (0).Note that (11) strictly predicts a logarithmic slope of unity, s = dX/d ln([ Ca ]) = 1.A similar equation was derived in [18].Figure 4A demonstrates an absence of divalent blockade for LESWAS in markedcontrast with the strong blockade for LEDWAS mutants, which is well-fitted bythe Langmuir isotherm (11) with a threshold value IC = 5 µ M ( s = 1 ± . IC value can in principle be connected to Q f [11] but it will require betterknowledge of the SF dimensions and will be a target of future research.Figure 4B shows that E rev for LEDWAS mutant starts from the same 50mM(the value measured for a 100mM Na + bath) as LESWAS but, from the pointwhere the current starts to increase with growth of [ Ca ] ( ≈ bath. It implies that,similarly to the L-type calcium channel, AMFE in the LEDWAS mutant involvesthe substitution of the sodium current by the calcium one. Equation (11) is alsoapplicable to the drug-driven blockade of bacterial mutants [5].Taken together, the results described above provide some experimental vali-dation of the ICB model. In particular, they confirm the importance of Q f as adeterminant of NaChBac ionic valence selectivity. Increasing the negative chargein the SF results in permeability for divalent cations and it leads to phenomenasuch as divalent blockade of the Na + current and AMFE. Moreover, the close fit-ting of the current decay by (11) confirms one of the main ICB results, viz. that the SF occupancy is described by a Fermi-Dirac distribution.Figure 5 illustrates diagrammatically the quasi-periodic { n } –sequence of multi-ion blockade/conduction modes described by (8) with growth of { n } , where Q f or P b increase, together with putative identifications of particular modes. The statewith Z = 0 represents ionic Coulomb blockade of the ions at the selectivity filter Fig. 5
Evolution of the Ca conduction mechanism with increasing absolute value of effectivefixed charge | Q f | , showing the Coulomb blockade oscillations of multi-ion conduction/blockadestates. The neutralized states Z n providing blockade are interleaved with resonant conductionstates M n . The | Q f | value increases from top to bottom, as shown. Green circles indicateCa ions, unfilled circles show vacancies (virtual empty states during the knock-on process).The right-hand column indicates the preliminary identifications of particular channels/mutantscorresponding to particular mechanisms. by image forces – as observed experimentally in LASWAS (see above) and also inartificial nanopores [13]. The first resonant point M corresponds to single-ion (i.e. { n } = ) barrier-less conduction, and can be related to the OmpF porin [10, 29].This state is followed by Z and M states describing double-ion knock-on andidentified with L-type calcium channels [9]. The three-ion resonance M can beidentified with the RyR calcium channels [30]. On a preliminary basis, NaChBacchannels can be identified with the Z blockade point, and their LEDWAS mutantwith the calcium-selective M resonant point. Further molecular dynamics simula-tions will be needed to resolve the observed difference between the nominal ( − e )and effective ( ≈ − e ) values of Q f for LEDWAS (see also [31]. Conclusions
We have reported the initial results of the first biological experiments undertakento test the predictions of the ICB model of ion channel conduction. In particu-lar, we used patch-clamp experiments to investigate Ca /Na + conduction and selectivity, AMFE, and ionic concentration dependences in the bacterial NaChBacchannel ( Q f = − e ) and in its charge-varied mutants with Q f = 0 and Q f = − e .The results are compared with earlier Brownian dynamics simulations of the per-meation process and with theoretical predictions of the ICB model, which we haveextended to encompass bulk concentration affects. onic Coulomb blockade and AMFE in NaChBac bacterial ion channels 11 We find that the ICB model provides a good account of both the experimen-tal (AMFE and valence selectivity) and the simulated (discrete multi-ion con-duction and occupancy band) phenomena observed in Ca channels, includingconcentration-related shifts of conduction/occupancy bands. In particular we haveshown that growth of Q f from − e to − e leads to strong divalent blockade of thesodium current by micromolar concentrations of Ca ions, similar to the effectsseen in calcium channels. The onset of divalent blockade (shape of the current-concentration curve) follows the Langmuir isotherm, consistent with ICB modelpredictions. Acknowledgements
The authors gratefully acknowledge valuable discussion with Igor Khovanov, CarloGuardiani and Aneta Stefanovska. The research was supported by the UK Engi-neering and Physical Sciences Research Council grant No. EP/M015831/1, “IonicCoulomb blockade oscillations and the physical origins of permeation, selectivity,and their mutation transformations in biological ion channels”.
Appendix 1. Generation, expression and measurements of NaChBacchannels R (cid:13) SDM Kit (NewEngland BioLabs Inc.) in accordance with the manufacturers instructions. Allmutations were confirmed by DNA sequencing prior to transfection of ChineseHamster Ovary (CHO) cells with TransIT-2020 (Mirus Bio). Transfected cells (ex-pressing GFP) were identified with an inverted fluorescence microscope (NikonTE2000-s) and their electrophysiological properties were determined 24–48 hoursafter transfection.Whole-cell currents were recorded using an Axopatch 200A (Molecular De-vices, Inc., USA) amplifier. Patch clamp signals were digitized using Digidata1322(Molecular Devices, Inc., USA) and filtered at 2 kHz. Patch-clamp electrodes werepulled from borosilicate glass (Kimax, Kimble Company, USA) and exhibited re-sistances of 23 MOhm. The shanks of the pipettes tip were coated with beeswaxin order to reduce pipette capacitance. The pipette (intracellular) solution con-tained (in mM): 120 Cs-methanesulfonate, 20 Na-gluconate, 5 CsCl, 10 EGTA,and 20 HEPES, pH 7.4 (adjusted by CsOH). Giga-Ohm seals were obtained in thebath (external) solution containing (in mM): 140 Na-methanesulfonate, 5 CsCl, 10HEPES and 10 glucose, pH 7.4 (adjusted by CsOH), in which Na-methanesulfonate then was subsequently replaced with Ca-methanesulfonate in order to vary Na + and Ca solution content (see main text). We used methanesulfonate salts in so-lutions to diminish the influence of endogenous chloride channels. Solutions werefiltered with a 0.22 mm filter before use. Osmolarity of all solutions was 280 mOsm(adjusted using sorbitol). Currentvoltage data were typically collected by recording responses to a con-secutive series of step pulses from a holding potential of −
100 mV at intervals of15 mV beginning at +95 mV. The bath solution was grounded using a 3M KClagar bridge. All experiments were conducted at room temperature.
References
1. B. Hille,
Ion Channels Of Excitable Membranes , 3rd edn. (Sinauer Associates, Sunderland,MA, 2001)2. J. Zheng, M.C. Trudeau (eds.),
Handbook of Ion Channels (CRC Press Taylor & FrancisGroup, Boca Raton, FL, 2015)3. B. Eisenberg, Physiol. (1), 28 (2013)4. L.X. Yue, B. Navarro, D.J. Ren, A. Ramos, D.E. Clapham, J. Gen. Physiol. (6), 845(2002)5. L. Tang, T.M.G. El-Din, T.M. Swanson, D.C. Pryde, T. Scheuer, N. Zheng, W.A. Catterall,Nature (7618), 117 (2016)6. C.E. Naylor, C. Bagn´eris, P.G. DeCaen, A. Sula, A. Scaglione, D.E. Clapham, B.A. Wal-lace, EMBO J. p. doi: 10.15252/embj.201593285 (2016)7. J. Payandeh, T.M.G. El-Din, T. Scheuer, N. Zheng, W.A. Catterall, Nature (7401),135 (2012)8. B. Corry, M. Thomas, J. Amer. Chem. Soc. (3), 1840 (2012)9. I.K. Kaufman, D.G. Luchinsky, R. Tindjong, P.V.E. McClintock, R.S. Eisenberg, Phys.Biol. (2), 026007 (2013)10. I.K. Kaufman, D.G. Luchinsky, R. Tindjong, P.V.E. McClintock, R.S. Eisenberg, Phys.Rev. E (5), 052712 (2013)11. I.K. Kaufman, P.V.E. McClintock, R.S. Eisenberg, New J. Phys. (8), 083021 (2015)12. M. Krems, M. Di Ventra, J. Phys. Condens. Matter , 065101 (2013)13. J. Feng, K. Liu, M. Graf, D. Dumcenco, A. Kis, M. Di Ventra, A. Radenovic, NatureMater. (8), 850 (2016)14. E. von Kitzing, in Membrane Proteins: Structures, Interactions and Models: Proc. 25thJerusalem Symposium on Quantum Chemistry and Biochemistry, Jerusalem, May 18-21,1992 , ed. by A. Pullman, J. Jortner, B. Pullman (Kluwer, Dordrecht, 1992), pp. 297–31415. R.S. Eisenberg, in
New Developments in Theoretical Studies of Proteins , ed. by R. Elber(World Scientific, Singapore, 1996), pp. 269–35716. J. Zhang, A. Kamenev, B.I. Shklovskii, Phys. Rev. E , 051205 (2006)17. J.L. Liu, R. Eisenberg, J. Phys. Chem. B , 12051 (2013)18. J.L. Liu, B. Eisenberg, J. Chem. Phys. (7), 075102 (2014)19. C.W.J. Beenakker, Phys. Rev. B (4), 1646 (1991)20. B. Corry, T.W. Allen, S. Kuyucak, S.H. Chung, Biophys. J. (1), 195 (2001)21. B. Roux, T. Allen, S. Berneche, W. Im, Quart. Rev. Biophys. (1), 15 (2004)22. W. Nonner, D. Gillespie, D. Henderson, B. Eisenberg, J. Phys. Chem. B (27), 6427(2001)23. D. Krauss, B. Eisenberg, D. Gillespie, Eur. Biophys. J. , 775 (2011)24. D.G. Luchinsky, W.A.T. Gibby, I. Kaufman, D.A. Timucin, P.V.E. McClintock, arXivpreprint arXiv:1604.05758 (2016)25. P.H. Nelson, J. Chem. Phys. , 165102 (2011)26. Y.M. Sun, I. Favre, L. Schild, E. Moczydlowski, J. Gen. Physiol. (6), 693 (1997)27. P.G. DeCaen, Y. Takahashi, T.A. Krulwich, M. Ito, D.E. Clapham, eLife , e04387 (2014)28. W.A. Sather, E.W. McCleskey, Ann. Rev. Physiol. , 133 (2003)29. H. Miedema, A. Meter-Arkema, J. Wierenga, J. Tang, B. Eisenberg, W. Nonner, H. Hektor,D. Gillespie, W. Meijberg, Biophys. J. (5), 3137 (2004)30. D. Gillespie, Biophys. J. (4), 1169 (2008)31. I.K. Kaufman, D.G. Luchinsky, W.A.T. Gibby, P.V.E. McClintock, R.S. Eisenberg, J.Stat. Mech. (5), 054027 (2016)32. D.J. Ren, B. Navarro, H.X. Xu, L.X. Yue, Q. Shi, D.E. Clapham, Science294