Insight into the selectivity and gating functions of Streptomyces lividans KcsA

Abstract

Streptomyces lividans KcsA is a 160-aa polypeptide that oligomerizes to form a tetrameric potassium channel. The three-dimensional structure of the polypeptides has been established, but the selectivity and gating functions of the channel remain unclear. It has been shown that the polypeptides copurify with two homopolymers, poly[(R)-3-hydroxybutyrate] (PHB) and inorganic polyphosphate (polyP), which have intrinsic capacities for cation selection and transport. PHB/polyP complexes are highly selective for divalent cations when pH is greater than the pK₂ of polyP (≈6.8), but this preference is lost when pH is ≤pK₂. It is postulated that KcsA polypeptides attenuate the divalent negative charge of the polyP end unit at physiological pH by strategic positioning of two C-terminal arginines. Here we mutate one or both of the C-terminal arginines and observe the effects on channel selectivity in planar lipid bilayers. We find that channels formed by KcsA polypeptides that retain a single C-terminal arginine remain highly selective for K⁺ over Mg²⁺, independent of medium pH; however, channels formed by KcsA polypeptides in which both C-terminal arginines have been replaced with neutral residues are selective for Mg²⁺ when pH is >7 and for K⁺ when pH is <7. Channel gating may be triggered by changes in the balance between the K⁺ polyP⁻ binding energy, the membrane potential, and the gradient force. The results reveal the importance of the C-terminal arginines to K⁺ selectivity and argue for a supramolecular structure for KcsA in which the host polypeptides modify the cation preference of a guest PHB/polyP complex.

Streptomyces lividans KcsA is a 160-aa polypeptide that oligomerizes to form a tetrameric potassium channel. The three-dimensional structure of the polypeptides has been well established. X-ray crystallographic analysis of the membrane-spanning portion of KcsA (residues 23–119) (1) indicates that four identical subunits form an inverted teepee surrounding a large central cavity leading to a narrow pore at the extracellular end. Site-directed spin labeling and EPR spectroscopy (2, 3) and NMR studies (4) indicate that the N terminus of KcsA forms a short amphipathic α-helix, whereas the C-terminal domains form a helical bundle that extends ≈50 Å into the cytoplasm.

Functional studies of KcsA are less clear. Most studies probing the mechanism of KcsA selectivity have been based on the concept of hydrated cations entering a central water-filled cavity through openings between the polypeptide helices at the inner membrane surface and exiting through a narrow “selectivity” pore at the extracellular end of the channel as the K⁺ hydration shell is replaced by carbonyl oxygen atoms of the pore-lining residues (3, 5–7). Gating is said to be governed by low intracellular pH of <5 (8–10), which effects movements of the inner helices that change the diameter of the permeation pathway (11–13). Yet intracellular pH levels are tightly regulated in S. lividans as in most organisms, so that protons are not likely to serve as natural activators of KcsA. Indeed, the widely held view of KcsA as a pH-dependent, proton-activated channel emanates from its failure to open at physiological pH when incorporated into planar lipid bilayers between symmetric K⁺ solutions (9, 10). In vivo, KcsA functions with a strong outward K⁺ gradient, and appropriately the channel opens readily at physiological pH when provided with an outward K⁺ gradient >2.5:1 (14). Both single-channel conductance and open probability (Po) increase with the magnitude of the K⁺ gradient (15).

Here we examine a different mechanism of selectivity and gating, based on the finding that KcsA copurifies with two ubiquitous homopolymers, poly[(R)-3-hydroxybutyrate] (PHB) (16, 17) and inorganic polyphosphate (polyP) (18, 19), that have a demonstrated capacity for cation selection and transport (20–22). PHB, a flexible, amphiphilic polyester, has solvent properties; polyP, a chain of negatively charged phosphoryl residues, selects, binds, and conducts cations. Studies indicate that PHB is covalently attached, presumably at its CoA ester end, to each KcsA monomer whereas polyP is held within the KcsA tetramer by noncovalent interactions (23). The presence of the polyanion, polyP, within KcsA tetramers was first signaled by the large difference between the theoretical pI of the polypeptides (10.3, EXPASY) and the experimental pI of the tetramer (6.5– 7.5). In contrast, the pI of the monomer was in agreement with the theoretical value (> 10), indicating that polyP is lost when the tetramer dissociates. The identity of polyP was confirmed by its metachromatic reaction to o-toluidine blue stain on an SDS/PAGE gel (24), its degradation on treatment with Saccharomyces cerevisiae polyphosphatase X (scPPX1) (25), and an enzymatic assay which demonstrated its ability to convert [¹⁴C]ADP to [¹⁴C]ATP in the presence of polyphosphate kinase (PPK) (26). PHB was detected in both tetrameric and monomeric forms by Western blot analysis and chemical assay (27).

PHB/polyP complexes form cation-selective ion channels in bacterial and mitochondrial membranes (20–22). Both PHB and polyP have been found to be constituents of the human erythrocyte plasma membrane calcium pump (28) and outer membrane protein A of nontypeable Haemophilus influenzae (29). A signal characteristic of channels formed by PHB/polyP complexes is their shift in selectivity from divalent to monovalent cations when pH is decreased from ≈7.5 to ≈6.0 (30). This change in charge preference is attributed to the pK₂ of polyP which is ≈6.8. The channels strongly prefer divalent cations when pH is greater than the pK₂ of polyP, because of the divalent negative charge of the polyP end unit and the stronger binding energy of divalent cations, but this preference is lost when pH equals pK₂, and the channels become increasingly selective for monovalent cations when pH is less than pK₂. Because physiological pH is strictly maintained when pH is >7, the premise is that selectivity for monovalent K⁺ is effected at the intracellular end of KcsA by strategic placement of the cytoplasmic C-terminal arginine residues of the polypeptides so as to attenuate the divalent negative charge of the polyP end unit.

Functional evidence for the participation of polyP in selectivity at the intracellular side of KcsA was reported by Zakharian and Reusch (31) who found that the selectivity of channels formed in planar bilayers by recombinant KcsA, which had not been fully reconstituted, was sharply dependent on medium pH at the intracellular side, i.e., Mg²⁺ was preferred at when pH was >7 and K⁺ was preferred when pH was <7. However, after annealing at temperatures >26°C, the channels became strongly selective for K⁺ and insensitive to medium pH. It was argued that the cytoplasmic C-terminal segments became disordered during purification procedures, thereby exposing polyP to medium pH. Incubation >26°C allowed the C-terminal segments to relax into their native position in which the eight terminal arginine residues of the four polypeptides presumably surround the polyP end unit and attenuate its divalent negative charge, thus having an effect similar to a lowering of pH. If this interpretation is correct, mutation of the C-terminal arginines to neutral residues should make channel selectivity dependent on intracellular pH. On the other hand, if the current model is correct and the C-terminal domains are not even part of the ion conduction pathway (3), mutation of the C-terminal arginines, 70–80 Å away from the narrow “selectivity” pore, should have no significant effect on channel selectivity.

Results

Preparation of Wild-Type and Mutant KcsAs.

The cytoplasmic C-terminal strands (residues 112–160) of the four identical polypeptides of wild-type KcsA have the sequence WFVGRE-QERRGHFVRHSEKAAEEAYTRTTRALHERFDRLER-MLDDNRR (pI 9.9). To determine the influence of the two C-terminal arginines (RR) of the polypeptides on channel selectivity, site-directed mutagenesis was used to prepare mutants in which one or both C-terminal arginines were converted to neutral residues (159V; 159V:160L or 159N:160N). Each protein was overexpressed in Escherichia coli cells and purified by Ni agarose chromatography (see Materials and Methods). The incidence of PHB and polyP in the purified proteins was confirmed. PHB was detected in both tetramers and monomers by Western blot analysis using anti-PHB IgG, and the amount was estimated by chemical assay (27). Because there are presently no standards for protein-conjugated PHB, this value was estimated relative to granule PHB. As a result, the assay provides relative concentrations of PHB but may not accurately reflect the true in vivo concentration. PolyP was detected in tetramers by o-toluidine blue stain (24). The length of polyP was estimated by acrylamide gel electrophoresis (32), and the identity of polyP was confirmed by its complete degradation by treatment with scPPX1 (25). No measurable differences were found in the concentrations of PHB and polyP in wild-type and mutant proteins; each contained 12 ± 5 residues per monomer unit of PHB and 60 ± 12 residues of polyP.

Representative Current Records.

Each protein was incorporated into micelles of tetraoxyethylene mono-n-octylether (C₈E₄) micelles. The channels were fully reconstituted by heating at 37°C for at least 2 h, and they were then incorporated into bilayers composed of synthetic 1-palmitoyl, 2-oleoyl phosphatidylcholine (POPC), synthetic 1-palmitoyl, 2-oleoyl phosphatidylethanolamine (POPE), and synthetic 1-palmitoyl,2-oleoyl phosphatidylglycerol (POPG) (3:3:1), formed across an aperture of ≈150 μm between gradient solutions of KCl (200 mM cis, 20 mM trans), 20 mM Hepes (pH 7.4) at 22°C. Representative outward current records are shown at +100 mV in Fig. 1. The wild-type (RR) and the single mutant (VR) display the same high conductance (220 ± 20 pS), and differ only in that the wild-type has a larger open probability (0.9 vs. 0.7). The mutants in which both terminal arginines were changed to neutral residues, either hydrophobic VL or hydrophilic NN, also display high open probabilities (0.9 and 0.7, respectively); however, they exhibit much lower conductance (30 ± 4 pS).

Fig. 1.

Representative current records of wild-type (RR) and mutant KcsA channels R159V mutant (VR), R159V,R160L mutant (VL), and R159N,R160N mutant (NN) at +100 mV in planar bilayers of synthetic POPC, POPE, and POPG (3:3:1) between gradient solutions of KCl (200 mM cis, 20 mM trans), 5 mM MgCl₂, 20 mM Hepes, and Tris (pH 7.4).

Selectivity for K⁺ over Mg²⁺ at Physiological pH.

The major physiological cations with outward gradients are K⁺ and Mg²⁺. The selectivity of the wild-type and mutant KcsAs for K⁺ over Mg²⁺ was examined at physiological pH by determining the effect of Mg²⁺ on the reversal potential for K⁺, i.e., the potential at which the current changes its direction. The Nernst theoretical potential for K⁺ (33), representing perfect selectivity for K⁺ under the conditions used in the measurements, is −54 mV. The channels were each incorporated into bilayers of synthetic POPC, POPE, and POPG (3:3:1) between 10:1 gradient solutions of KCl as above at pH 7.4. The reversal potentials were determined from current/voltage (I/V) relations in the absence of Mg²⁺ and then after addition of symmetric low concentrations of Mg²⁺ (1–5 mM).

A representative IV curve for the above solutions containing 5 mM Mg²⁺ is shown in Fig. 2. The insert shows the region in which the current reverses direction. The reversal potentials of wild-type KcsA (RR) and single mutant VR are essentially at the Nernst potential (arrow); the two double mutants (VL and NN) display a reversal potential of approximately −27 mV. The reversal potentials for wild-type (RR) and double mutant VL are also shown in voltage ramps from −100 mV to +100 mV under the same conditions in Fig. 3. The Nernst potential (double arrow) and experimental reversal potential (single arrow) for wild-type RR are essentially the same, whereas the experimental reversal potential for VL is shifted to more positive values.

Fig. 2.

Representative current/voltage (I/V) curve. Recombinant wild-type (RR; ■), R159V mutant (VR; ○), R159V,R160L mutant (VL; ▴), and R159N,R160N mutant (NN; ▿) channels were each incorporated in planar bilayers of synthetic POPC, POPE, and POPG (3:3:1) between gradient solutions of KCl (200 mM cis, 20 mM trans), 20 mM Hepes, and Tris (pH 7.4), with symmetric 5 mM Mg²⁺. (Inset) Expanded version of the region near the reversal potential. RR, solid line; VR, dashed line; VL, dotted–dashed line; NN, dotted line.

Fig. 3.

Voltage scans of wild-type RR and double mutant VL (R159V,R160L) from −100 mV to +100 mV. Proteins were each incorporated in planar bilayers between gradient K⁺ solutions as in Fig. 2 (pH 7.4), with symmetric 5 mM Mg²⁺.

The reversal potentials as a function of [Mg²⁺], determined from current/voltage (I/V) relations at pH 7.4, are summarized in Fig. 4. As above, there is no significant change in the reversal potential for wild-type KcsA (RR) and single mutant VR on addition of 1–5 mM Mg²⁺ to both sides of the bilayer, indicating impermeance of Mg²⁺. However, the double mutants, VL and NN, both exhibit sharp increases in reversal potential with incremental addition of Mg²⁺, indicating a strong preference for Mg²⁺. The results indicate that the large outward currents observed for wild-type (RR) and mutant VR (Fig. 1) are K⁺ currents, whereas the small outward currents observed for the double mutants, VL and NN, are mainly Mg²⁺ currents.

Fig. 4.

Selectivity for K⁺ as a function of [Mg²⁺]. Recombinant wild-type (RR; ■), R159V mutant (VR; ○), R159V,R160L mutant (VL; ▴), and R159N,R160N mutant (NN; ▿) were each incorporated in planar bilayers between gradient K⁺ solutions as in Fig. 2 (pH 7.4). [Mg²⁺] was as designated. Reversal potentials were determined from I/V relationships (see Fig. 2). Arrow shows the Nernst theoretical potential for K⁺. Errors in the measurements were ±5%.

Selectivity for K⁺ over Mg²⁺ as a Function of pH.

For studies of the influence of pH on selectivity, the channels were each incorporated into bilayers of synthetic POPC, POPE, and POPG (3:3:1) between a 10:1 gradient solution of KCl as above with symmetric 5 mM MgCl₂, 20 mM KHepes, and the pH was varied in steps from 7.4 to 5.0. Reversal potentials were determined from I/V relationships as above. As shown in Fig. 5, the reversal potentials of wild-type (RR) and single mutant VR remain near the Nernst potential for K⁺ as pH is decreased from 7.4 to 5.0, indicating the strong selectivity of the channels for K⁺ is independent of pH. However, the double mutants, VL and NN, which exhibited a strong preference for divalent Mg²⁺ at physiological pH (Figs. 2–4), exhibit sharp decreases in reversal potentials as pH is decreased, approaching the Nernst potential for K⁺ at a pH of ≈6.0. This indicates a shift in preference from divalent to monovalent cations as pH is decreased.

Fig. 5.

Selectivity for K⁺ as a function of pH. Recombinant wild-type (RR; ■), R159V mutant (VR; ○), R159V,R160L mutant (VL; ○▴), and R159N,R160N mutant (NN; ○▿) were each incorporated in planar bilayers between gradient K⁺ solutions, 5 mM MgCl₂, as in Fig. 2. pH was as designated. Reversal potentials were determined from I/V relationships. Arrow shows the Nernst theoretical potential for K⁺. Errors in the measurements were ±5%.

Discussion

Our results point to the cytoplasmic C-terminal face of KcsA as the site of cation selectivity. The remarkable loss of K⁺ selectivity following the mutation of the two C-terminal arginines to neutral residues, either hydrophobic VL or hydrophilic NN, cannot be explained by the current model which holds that selection takes place at the narrow pore at the extracellular end of the vestibule, which is ≈30 Å from the intracellular side of the membrane and ≈70–80 Å from the C-terminal residues. Moreover, the involvement of polyP in selectivity is strongly indicated by the fact that the midpoint of the switch in cation preference of KcsA mutants VL and NN is the pK₂ of polyP; both exhibit a preference for divalent cations when pH is greater than the pK₂ of polyP and for monovalent cations when pH is less than the pK₂ of polyP. PHB/polyP channels undergo a shift in selectivity from divalent (Ca²⁺) to monovalent (Na⁺) in the same pH range (27). Physiological pH is maintained moderately above the pK₂ of polyP, thus only a small decrease in local pH is required to shift the charge preference of polyP from divalent to monovalent.

Once the supramolecular composition of the channel is recognized, its selectivity is readily understood as interplay between the three polymers and the intracellular and extracellular extant cations. In this perspective, the narrow pore at the external end of the vestibule plays a crucial role in forming a barrier to prevent influx of cations with inward gradients, such as Ca²⁺ and Na⁺, from the extracellular fluid, whereas selectivity at the intracellular side is carried out at the mouth of the channel by cooperative interactions between the three polymers. Potassium ions are selected, dehydrated and held along the polyP backbone chain ready for export through the narrow pore in response to a stimulus. The outward gradient provides the impetus for K⁺ flow but this is countered by strong ionic bonds between K⁺ and polyP and the negative membrane potential. “Gating” occurs when the equilibrium between these forces is disturbed.

PHB/polyP complexes have an intrinsic capacity for cation selection and transport. It seems reasonable to assume that, over time, polypeptides evolved to associate with and shape the local environment of the complexes so as to enable the transmembrane movement of diverse cations. The prokaryotic KcsA may well be an example of how polypeptide composition and structure may be fashioned to harness the conductive potential of PHB/polyP for the controlled export of K⁺ from the cytoplasm under physiological conditions.

Materials and Methods

Preparation of KcsA Mutants.

Wild-type KcsA, His-tagged at the C-terminal and cloned into pQE60, was a gift from C. Miller (Brandeis University, Waltham, MA). All mutants were made by using the Quik-Change site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions.

Purification of S. lividans KcsA Wild-Type and Mutants.

pQE60 plasmids were transformed into E. coli BL21 (Novagen), overexpressed by addition of isopropyl β-d-thiogalactopyranoside (IPTG) to a final concentration of 1 mM (Calbiochem, San Diego, CA) and purified by Ni-affinity chromatography as described (23). The proteins when unheated formed single bands at ≈65 kDa on SDS/PAGE gels, corresponding to the tetrameric form, and they were converted to the monomeric form, ≈19 kDa, when heated in 2% SDS.

Assay for PHB.

The presence of PHB in KcsA wild-type and mutants was demonstrated by Western blot analysis using anti-PHB IgG as described (23). The identity of PHB was confirmed by chemical assay as described by Huang and Reusch (27).

Assay for PolyP.

The presence of polyP in wild-type KcsA and KcsA mutants was probed by labeling during overexpression with ³²P-PO₄²⁻. ³²P-Labeled KcsA was incubated with proteinase K (20 mg/ml), and protein was extracted with 1/2 volume phenol:chloroform (1:1) and then chloroform. The aqueous layer, containing ³²P-polyP was concentrated and subjected to electrophoresis on a 15% acrylamide gel (19:1; acrylamide:bis) using Tris/Borate/EDTA (90/90/2.7 mM, pH 8.3) (32). Xylene cyanol and bromphenol blue, which migrate to the same position as 72 and 45 polyP units, respectively, were used as standards. The identity of polyP was confirmed by observing the complete degradation of the polymer upon addition of scPPX1 (25) (courtesy of A. Kornberg, Stanford University, Stanford, CA).

Planar Lipid Bilayer Measurements.

The KcsA wild-type and mutant proteins were each reconstituted into liposomes by incubation at 30°C for >2 h in a micellar solution composed of 20 mM tetraoxyethylene mono-n-octylether (C₈E₄; Sigma–Aldrich, St. Louis, MO) and a mixture of synthetic POPC, synthetic POPE, and synthetic POPG (3:3:1) (Avanti Polar Lipids, Alabaster, AL). Planar lipid bilayers were formed from a solution of POPC:POPE:POPG (3:3:1) in n-decane (Sigma–Aldrich). The lipid solution was used to paint a bilayer on an aperture of ≈150 μm diameter in a Delrin cup (Warner Instruments, Hamden, CT) between aqueous bathing solutions as described in Results. All salts were ultrapure (>99%) (Sigma–Aldrich). Fluctuations of junction potential, estimated with the JPCALC of the pClamp software, were 0.5–0.8 mV. Bilayer capacitances were in the range of 25–50 pF. After the bilayers were formed, ≈1 μl of the KcsA micellar solution (diluted to 20 μg/ml) was added to the cis compartment with gentle stirring.

Recording and Data Analysis.

Unitary currents were recorded with an integrating patch-clamp amplifier (Axopatch 200A; Axon Instruments, Foster City, CA). The cis compartment (voltage command side) was connected to the CV 201A head stage input, and the trans compartment was held at virtual ground via a pair of matched Ag-AgCl electrodes connected to the solutions by an agar bridge containing 3 M KCl. Currents through the voltage-clamped bilayers (background conductance <1–2 pS) were low-pass-filtered at 10 kHz (−3-dB cutoff, Bessel type response) and recorded after digitization through an analog-to-digital converter (Digidata 1322A; Axon Instruments).

Data were filtered through an eight-pole Bessel filter (902LPF; Frequency Devices, Haverhill, MA) and digitized at 1 kHz by using pClamp9 software (Axon Instruments). Singe-channel conductance events were identified automatically and analyzed by using Clampfit9 software (Axon Instruments). The data were averaged from ≈200 independent recordings. Nernst potentials were calculated using ion activities. Permeability ratios were determined from the Goldman Hodgkin–Katz voltage equation (33).

Temperature Studies.

For studies above room temperatures, a Teflon cuvette was seated in a special outer chamber made of a polymer/graphite mixture (Dagan Instruments, Minneapolis, MN). The chamber was fitted on a conductive stage containing a pyroelectric heater/cooler. Deionized water was circulated through this stage to remove the heat generated. The pyroelectric heating/cooling stage was driven by a temperature controller (HCC-100A; Dagan Instruments). The temperature of the bath was monitored constantly with a thermoelectric monitor in the trans side of the cuvette.

Abbreviations

PHB

poly[(R)-3-hydroxybutyrate]

polyP

inorganic polyphosphate

C₈E₄

tetraoxyethylene mono-n-octylether

POPC

1-palmitoyl, 2-oleoyl phosphatidylcholine

POPE

1-palmitoyl, 2-oleoyl phosphatidylethanolamine

POPG

1-palmitoyl, 2-oleoyl phosphatidylglycerol

scPPX1

Saccharomyces cerevisiae polyphosphatase X.