A codon-optimized version of the I131C stable mutant (Wang et al., 2009) of KirBac1.1 was custom synthesized (GenScript Biotech, Rijswijk, Netherlands) to reduce its guanine-cytosine content and subcloned into pET-28a. Protein sequence is:

MNVDPFSPHS¹⁰SDSFAQAASP²⁰ARKPPRGGRR³⁰IWSGTREVIA⁴⁰YGMPASVWRD⁵⁰LYYWALKVSW⁶⁰PVFFASLAAL⁷⁰FVVNNTLFAL⁸⁰LYQLGDAPIA⁹⁰NQSPPGFVGA¹⁰⁰FFFSVETLAT¹¹⁰VGYGDMHPQT¹²⁰VYAHAIATLE¹³⁰CFVGMSGIAL¹⁴⁰STGLVFARFA¹⁵⁰RPRAKIMFAR¹⁶⁰HAIVRPFNGR¹⁷⁰MTLMVRAANA¹⁸⁰RQNVIAEARA¹⁹⁰KMRLMRREHS²⁰⁰SEGYSLMKIH²¹⁰DLKLVRNEHP²²⁰IFLLGWNMMH²³⁰VIDESSPLFG²⁴⁰ETPESLAEGR²⁵⁰AMLLVMIEGS²⁶⁰DETTAQVMQA²⁷⁰RHAWEHDDIR²⁸⁰WHHRYVDLMS²⁹⁰DVDGMTHIDY³⁰⁰TRFNDTEPVE³¹⁰PPGAAPDAQA³²⁰FAAKPGEGDA³³⁰RPVLVPRGSR³⁴⁰SHHHHHH³⁴⁷

The C-terminal hexahistidine tagged KirBac1.1 channel was overexpressed in BL21 (DE3)-competent cells (Molina et al., 2004) with the following modifications. E. coli competent cells were transformed with the above construct following standard heat-shock procedures and plated overnight on LB agar. A single colony was picked up and grown overnight at 30°C in 100 mL of LB medium. This culture was diluted into 2 L of 2xYT medium and grown at 37°C to exponential phase (absorbance at 600 nm of ∼0.6). Protein expression was induced by 0.5 mM isopropyl β-D-thiogalactopyranoside at 30°C for 3 h. Cells were pelleted and suspended in 100 mL of buffer (20 mM HEPES, pH 7.5, 0.45 M sucrose) containing one EDTA-free protease inhibitor cocktail tablet (Roche, Madrid, Spain) and kept on ice for 1 h. The mixture was sonicated on an ice bath using a Branson probe-type apparatus and centrifuged for 45 min at 100,000 × g. Membrane proteins in this crude membrane pellet were solubilized in 40 mL of 20 mM HEPES, pH 7.5, 100 mM KCl, 50 mM imidazole and 10 mM n-dodecyl β-D-maltoside (DDM; Calbiochem, Madrid, Spain) for 2 h at room temperature. After centrifugation of insoluble remains (45 min at 100,000 × g), the supernatant was incubated with Ni²⁺- Sepharose 6 fast flow resin (Cytiva, Madrid, Spain) overnight at 4°C, placed into a column, and washed with 20 mM HEPES, pH 7.5, 100 mM KCl, 50 mM imidazole, and 5 mM DDM, until the absorbance at 280 nm was <0.01. The gel-bound protein was eluted using 15 mL of the previous buffer containing 500 mM imidazole. Protein concentration was routinely determined from the absorbance at 280 nm and always given in terms of KirBac1.1 monomers, using a molecular weight of 38.8 kDa and a molar extinction coefficient of 51,350 M⁻¹cm⁻¹ (Pace and Scholtz, 1997). The purified protein batches were checked by SDS-PAGE and BN-PAGE (Giudici et al., 2013).

KirBac1.1 was reconstituted in asolectin liposomes for the activity assays, as previously reported (Poveda et al., 2019). The required amount of asolectin (Soybean, type II-S, Sigma, Madrid, Spain) was dissolved in chloroform:methanol (2:1, by volume) and the solvents removed using a rotary evaporator and vacuum. The dried lipid film was resuspended at 10 mg/mL in 10 mM HEPES, pH 7.0, 100 mM KCl and stored in liquid nitrogen. Before use, defrosted lipid suspensions were diluted to 5 mg/mL, then vortexed and extruded through 100 nm filters to obtain large unilamellar vesicles. DDM solubilized protein channels at approximately 1 mg/mL were added drop by drop to the lipid solution while being vortexed to give a lipid-to-protein ratio of 50:1 by weight. Bio-Beads SM-2 (Bio-Rad laboratories, Madrid, Spain) were added and the suspension was rocked overnight at 4°C to remove detergent. After discarding the Bio-Beads, the reconstituted liposome suspensions were collected by centrifugation for 65 min at 100,000 × g (TFT-70; Beckman, California, United States) and finally suspended in 10 mM HEPES, pH 7.0, 100 mM KCl. Samples were stored at −80°C (Giudici et al., 2022).

For patch-clamp experiments, 3 µL of the above-reconstituted liposomes were placed on a clean glass slide and dried overnight in a desiccator chamber at 4°C and then rehydrated with 20 μL of 10 mM HEPES (pH 7.0), yielding multilamellar giant liposomes after a few hours of rehydration. A volume of 3–6 µL of the resulting giant liposome suspension were deposited onto a 3.5 cm Petri dish and mixed with 3 mL of buffer, the bath solution in the electrical recordings. Four different buffers were used either to fill the patch electrodes (pipette solution) or the bath solution, 10 mM MES pH 4 or 10 mM HEPES pH 7, containing either 200 mM of NaCl or KCl. Giga seals were formed on giant liposomes with borosilicate microelectrodes (Sutter Instruments, California, United States) of 3–5 MΩ open resistance. Standard inside-out patch-clamp recordings were performed using an EPC-9 (Heka Electronic, Lambrecht/Pfalzt, Germany) amplifier, at a gain of 50 mV/pA. Recordings were filtered at 1 kHz with an 8-pole Bessel filter. The holding potential was applied to the inside of the patch pipette, and the bath was maintained at virtual ground. An Ag−AgCl wire was used as the reference electrode. The data was analyzed with Clampfit 10.3 (Molecular Devices, Axon Instruments, San José, CA, United States). All measurements were made at room temperature (24°C). Reversal potentials were determined under bi-ionic conditions in order to obtain the permeability ratio of Na⁺ relative to K⁺, P_Na ⁺/P_K ⁺, from the Goldman–Hodgkin–Katz equation (Hille, 2001): E r e v = R T F ln P K + K + 0 + P N a + Na + 0 P K + K + i + P N a + Na + i (1) where E _rev is the calculated reversal potential (zero-current potential), R the gas constant, T the temperature (297 K), F the Faraday constant, [K⁺]_o and [K⁺]_i are the extracellular and intracellular K⁺ concentrations, respectively, and [Na⁺]_o and [Na⁺]_i refer to extracellular and intracellular Na⁺ concentrations, respectively.

Liquid junction potentials between the pipette and bath solutions were calculated by using Clampex 10.3 (Molecular Devices, Axon Instruments, San José, CA, United States) and routinely corrected. To measure the reversal potential of the channel currents under bi-ionic conditions, the holding potential was stepped to different voltages (10 mV steps), and the null current potential for each seal was the x-axis intersection point of the plotted I/V values.

KirBac1.1 reconstituted in asolectin liposomes in 10 mM HEPES pH 7, either with 100 mM KCl or NaCl, was incubated for 15 min at room temperature with a stock solution of the fluorescent probe 9-Amino-6-chloro-2-methoxyacridine (ACMA) in methanol, at a final probe concentration of 100 µM. Samples were diluted 1:160 with N-methyl-d-glucamine (NMDG) buffer at different pHs in a 5 × 5 mm quartz cuvette to create a cation gradient. These NMDG buffers are composed of 10 mM HEPES for pH 7, and 10 mM Succinic acid for pHs 4, 4.5, 5, 5.5, 6, 6.5, all of them with 200 mM NMDG⁺ acting as a non-conductive cation. All experiments were conducted at 25°C using a PicoQuant F300 spectrofluorometer with excitation and emission wavelengths of 410 nm and 480 nm, respectively. Fluorescence reading was taken every second for about five minutes to obtain a stable baseline (F _B ) before the addition of carbonyl-cyanide m-chlorophenylhydrazone (CCCP) at a final concentration of 2 μM, prepared from a stock in DMSO, to initiate the ion flux. CCCP is a proton ionophore which enables the entry of protons to counter the efflux of ions conducted by KirBac1.1. This flux of protons is then monitored by the membrane-permeable pH-sensitive dye ACMA, which at neutral pH has bright fluorescence but is quenched upon protonation and is no longer membrane-permeable. Fluorescence was monitored for about 10–15 min, reading every second to get a stable fluorescence (F _C ) before the addition of valinomycin (or gramicidin), which provides a channel-independent path for K⁺ (or K⁺ and Na⁺) conduction. Both ionophores were used at a 10 nM concentration prepared from a stock solution in ethanol, to determine minimal fluorescence (F _V ). Normalized fluorescence was calculated using: F N = F − F V / F B − F V (2) and the % relative K⁺ flux was calculated as: % f l u x = 100 x F C − F V / F B − F V (3)

%flux vs. pH curves were fitted with a Hill equation after transforming pH into H⁺ concentration: % f l u x = F M − F M x H + n k n + H + n (4) where [H⁺] is the concentration of H⁺, F _m is the maximum flux, n is the Hill coefficient, and k is the H⁺ concentration at which half of the maximum flux is reached, from which pH _0.5 is calculated (pH _0.5 = -log(k)).

Thermal denaturation of KirBac1.1 channel was monitored as described previously (Renart et al., 2010) using a Varian Cary Eclipse spectrofluorometer to record the temperature dependence of the protein intrinsic emission fluorescence at 340 nm after excitation at 280 nm. In these experiments, the protein was diluted to 1 µM concentration in either 20 mM HEPES buffer, pH 7.0, containing 5 mM DDM or 10 mM Succinic acid buffer, pH 4.0, containing 5 mM DDM and always in the presence of a fixed amount of the opposite cation being tested. The starting stock is always 20 mM HEPES buffer, pH 7.0, 5 mM DDM and 100 mM KCl. In order to have the protein in Succinic acid buffer or NaCl, a previous dialysis is performed. For the cation titrations, aliquots from stock solutions of either NaCl or KCl were added to the samples prior to the thermal denaturation recordings to provide the desired final cation concentrations. The midpoint temperature of dissociation and unfolding of the tetramer (tm, in Celsius) was calculated from the thermal denaturation curves by fitting a two-state unfolding model to the data (Renart et al., 2024). The dissociation constants of the KirBac1.1-cation complexes (K_D) can be estimated from: ∆ T m T m = T m − T m 0 T m = R T m 0 ∆ H 0 ln 1 + L K D (5) where T _m and (T _m ) ₀ refer to the denaturation temperature (in Kelvin) for the protein in the presence and absence of ligand, respectively, R is the gas constant, and ΔH ₀ is the enthalpy change upon protein denaturation in the absence of ligand.

Fluorescence emission spectra were taken on a Varian Cary Eclipse spectrofluorometer using 0.5 cm path-length quartz cuvettes, as described in (Renart et al., 2006). The samples were excited at 280 nm, and the emission was recorded between 300 and 400 nm in 1 nm increments.

Changes in the fluorescence emission spectrum of KirBac1.1 where quantified from the intensity-weighted average emission wavelength < λ >: < λ > = ∑ i λ i I i ∑ i I i (6) where I _i is the fluorescence intensity measured at a wavelength λ _i .

< λ > vs. pH curves were fitted with a Hill equation as in the flux assay (Equation 4) but substituting ion flux by < λ >. This fitting was used again to calculate pH _0.5 .

Samples used in these experiments were prepared as in the thermal denaturation assays at a KirBac1.1 final concentration of 1 μM, in a buffer of 10 mM HEPES, pH 7.0, 150 mM KCl (or NaCl), 5 mM DDM, or 10 mM Succinic acid, (pH 4.0, 4.5, 5.0, 5.5, 6.0, 6.5), 150 mM KCl (or NaCl), 5 mM DDM. All measurements were made at room temperature (25°C).

Channel function in our purified KirBac1.1, reconstituted in asolectin lipid vesicles, has been assessed by two different procedures. First, we carried out a macroscopic assay of K⁺ flux by following fluorescence quenching (the ACMA/CCCP assay, as described in Methods). Figure 2A shows that, as expected for Kir channels, regardless of their prokaryotic or eukaryotic origin (Sepúlveda et al., 2015), channel opening is strongly pH-dependent. Figure 2B illustrates further such dependence, yielding an estimated pH_0.5 for channel opening of 5.5 ± 0.1, which is similar to that reported from Rb⁺ fluxes in this same channel (Enkvetchakul et al., 2004). Overall, these experiments indicate that our purified KirBac1.1 channel is active and that, as an experimental system, the closed or the open channel states may be favored by selecting the adequate pH conditions.

All the experiments reported in Figure 2 were conducted having K⁺ in the intravesicular compartment and a non-permeable cation, NMDG⁺, extravesicularly. Nonetheless, the same ACMA/CCCP assay can also be used to estimate cation selectivity by replacing K⁺ by Na⁺ inside the vesicles. In these conditions, if the KirBac1.1 channel were impermeable to Na⁺, no fluorescence quenching would be expected to occur. However, Figure 3 shows that Na⁺ efflux occurs at pH 7, although the process is less efficient than for K⁺. This suggests that KirBac1.1 is only partly selective for K⁺, although its SF is identical to that of KcsA, a channel markedly selective for K⁺ (LeMasurier et al., 2001; Montoya et al., 2017). Figure 3 also shows that at pH 4 no flux of neither K⁺ nor Na⁺ can be observed, indicating again that acidic pH favors the closed state of the channel, regardless of the ion used in the assays.

To study the gating of KirBac1.1 in more detail, our second approach to measuring its function was to use patch-clamp methods on patches excised from reconstituted giant liposomes (Giudici et al., 2022). At pH 7 and 200 mM K⁺ at both sides of the patches (symmetrical conditions), an irregular activity with different current levels was observed (Figure 4A), which is similar to observations reported by other authors using reconstituted membranes with a different lipid composition (Cheng et al., 2009). Likewise, at pH 7 and under symmetrical 200 mM Na⁺ concentrations, channel activity is also observed (Figure 4B), although current levels were lower and the recordings even more flickery than those observed in K⁺.

The relative selectivity of K⁺ versus Na⁺ was measured in membrane patches under asymmetric ionic conditions (200 mM K⁺ into the pipette solution and 200 mM Na⁺ in the bath). An intensity versus voltage (I/V) plot at pH 7 (Figure 5A) shows that outward K⁺ currents were slightly larger than Na⁺ inward currents. No signs of current rectification were observed because, as in many other Kir channels, rectification requires the presence of certain cofactors (Mg²⁺, spermidine and others) (Cheng et al., 2009) which are not present in our experimental system. In spite of the difficulties to measure current size in the recordings, a reversal potential of 21.5 mM was estimated (Equation 1), leading to a permeability ratio (P_K ⁺/P_Na ⁺) of 2.3. Such parameters are far from those estimated for truly selective K⁺ channels such as KcsA (LeMasurier et al., 2001; Montoya et al., 2017) and confirms the above observations from the ACMA/CCCP assays that the KirBac1.1 channel is only partly selective for K⁺.

At pH 4, most of the patches were silent, which is also in agreement with the ACMA/CCCP observations that the channels are inactive or closed at this pH. Nonetheless, channel electrical activity could still be observed in a percent of the patches under these conditions of acidic pH. Such activity is always fairly symmetric in terms of inward and outward current sizes and very flickery (Figures 5A, C). The I/V curves from such recordings yield a reversal potential of 0.8 mV and a relative permeability ratio near 1, indicating that at acidic pH, the channels that remain active are non-selective and able to conduct K⁺ or Na⁺ similarly.

The KirBac1.1 channel used in these studies contains a considerable number of tryptophan residues located at the cytoplasmic C-terminal protein domains (see Figure 1). Therefore, the channel is intrinsically fluorescent and we used such property to check spectroscopically whether there are detectable protein structural changes as a consequence of pH or in the presence of a given cation. All the following experiments were conducted with the purified protein in plain detergent (DDM) solutions. This eliminates both, the possible influence on the results caused by the presence of specific lipids (Singh et al., 2009; Amani et al., 2020) and excessive light scattering arising from the presence of lipid vesicles.

Figures 6A, B show that, regardless of whether the samples were prepared in a Na⁺ or K⁺ buffer, the fluorescence emission spectrum undergoes a small but reproducible blue shift when the pH swiches from pH 7 to 4. This suggests that pH causes structural changes in the cytoplasmic C-terminal region of the protein channel. The occurrence of conformational changes with pH was also suggested in previous FRET and NMR studies (Sadler et al., 2016; Amani et al., 2020). Interestingly, the fluorescence spectral shifts observed here, although small, allow to do pH titration experiments (Figure 6C), which clearly show a sigmoidal variation, very similar to that observed in the ACMA/CCCP assays of ion flux versus pH. The similarities between the pH_0.5’s obtained from application of the two techniques, monitoring very different experimental observables (the spectral shift and the ion flow), suggest that changes in the channel structure accompany the functional changes observed in the transition from the closed to the open channel states. Figure 6C also shows that, despite the different effects of Na⁺ or K⁺ on the SF described below, the fluorescence spectral changes were similarly observed regardless of the presence of either Na⁺ or K⁺ in the samples. This is not too surprising because, as stated above, the tryptophane reporter groups in KirBac1.1 are located at intracellular protein domains, while cation binding takes place at the distant channel SF (Kuo et al., 2003). In contrast to these observations on KirBac1.1, previous studies using KcsA, in which the reporter trytophane groups are adjacent to the selectivity filter, demonstrated an exquisite sensitivity of the protein fluorescence to the binding and the filling of the cation binding sites at the SF (Zhou and MacKinnon, 2003; Renart et al., 2006, 2010; 2019).

From the KcsA and NaK channels, we also learned that K⁺ and other cation’s binding to the SF, bridge together the subunits within the tetrameric channel structure, conferring a large increase in the protein thermal stability and protecting it against tetramer dissociation and partial unfolding (Renart et al., 2010; Giudici et al., 2022). Thus, we decided to use the thermal denaturation assay to check whether cation binding causes also changes in the stability of the KirBac1.1 channel. Figure 6D shows that in 100 mM K⁺, the interaction of K⁺ with the channel is much stronger at pH 7, with the open channel state, than with the closed channel at pH 4, with T_m values (the mid-point temperature of protein denaturation) increasing considerably (up to 10°C–20°C for different preparations). In contrast, in 100 mM Na⁺ (Figure 6E), the interaction between Na⁺ and the protein channel seems stronger with the closed channel state at pH 4.

Since the results from Figures 6D, E indicate important differences in the binding of K⁺ and Na⁺ to the open and closed states of the channel, titrations experiments were conducted with the two cations competing with each other to further illustrate the binding processes taking place. At pH 7, when the channel is in the open state, the addition of K⁺ increases considerably the protein thermal stability, regardless of the presence of increasing concentrations of Na⁺ (Figures 7A–C). As extensively proven with KcsA and NaK channels (Renart et al., 2010, 2023; Giudici et al., 2022), this suggests that K⁺ entering into the KirBac1.1 selectivity filter favors the “induced-fit” conformation in which the channel’s subunits get closer together, bridged by the coordinating bound cations, thus, resulting in an increased thermal stability. On the contrary, such increased stability associated to the “induced-fit” phenomenon is not observed in the Na⁺ titrations (Figures 7D–F), in which the only stabilization observed comes from the presence of the fixed K⁺ concentrations used in the experiments. This leads to the conclusion that, despite being a permeable species through the KirBac1.1 channel, Na⁺ does not bind to the selectivity filter, or if it does, it is unable to cause the “induced-fit” conformation observed upon K⁺ binding.

Titration experiments similar to those above were conducted also at pH 4, where the channel is in a closed state, with markedly different results. Under this condition, K⁺ continues to bind and stabilize the channel in the absence of Na⁺, but once Na⁺ is added at higher concentrations, there is an additional stabilization caused by this ion (Figures 8A–C). This suggests that in the closed channel state, both cations compete with each other to bind to the channel and cause a phenomenon similar to the “induced-fit” in the selectivity filter conformation, leading to thermal stabilization. It is tempting to speculate that this similar ability of the two cations to act on the selectivity filter, could be related to the observed lack of cation selectivity observed in the functional studies described above. Likewise, in the Na⁺ titrations shown in Figures 8D–F, it is observed that Na⁺ strongly stabilizes the channel’s closed state, similarly to the stabilization caused by K⁺ on the open channel state from above. Moreover, in the presence of competing identical concentrations of Na⁺ and K⁺ (Figure 8F) it is observed that Na⁺ is even more effective than K⁺ in stabilizing the protein, suggesting that the affinity of Na⁺ for binding to the closed channel could be even higher than that for K⁺ at this acidic pH.

In order to estimate the apparent dissociation constants for the above binding processes, Equation 5 as described in Methods, was fitted to the results from the titration experiments under competing conditions. Such fitting procedure has been used extensively in the past, mainly in studies on cation binding to the KcsA channel (Renart et al., 2017, 2024). Figure 9 shows the results from K⁺ titrations at pH 7, where the channel is in an open state. No results from Na⁺ titrations are included in the Figure because, as indicated above, the presence of Na⁺ does not result in any thermal stabilization of the channel under this condition. This Figure shows single exponential curves for K⁺ binding in the presence of three different concentrations of Na⁺, suggesting the existence of a single set of K⁺ binding sites, which are essentially unaffected by the simultaneous presence of Na⁺ and shows K⁺ dissociation binding constants in the millimolar range Table 1). Such values are similar to those low affinity constants exhibited by the KcsA channel for K⁺ binding, which were related to the conformation of the SF allowing K⁺ permeation (Renart et al., 2010).

Figure 10 shows the results from the cation titration experiments at pH 4, where the channel is in a closed state. The binding curves are now more complex, except in the absence of cation competition, where the exponentials are nicely defined. Still, the analysis through Equation 5 allows an estimation of the dissociation constants for the binding of the cations to single sets of cation binding sites in the millimolar range or higher (Table 1), although the affinity to bind Na⁺ seems always higher than that for K⁺ binding. Such estimations clearly illustrate that, as a consequence of competition between Na⁺ and K⁺ to bind to the closed channel, their binding affinities decrease progressively as the concentration of the competing cation increases.