Family of prokaryote cyclic nucleotide-modulated ion channels

Abstract

Cyclic nucleotide-modulated ion channels are molecular pores that mediate the passage of ions across the cell membrane in response to cAMP or GMP. Structural insight into this class of ion channels currently comes from a related homolog, MloK1, that contains six transmembrane domains and a cytoplasmic cyclic nucleotide binding domain. However, unlike eukaryote hyperpolarization-activated cyclic nucleotide-modulated (HCN) and cyclic nucleotide-gated (CNG) channels, MloK1 lacks a C-linker region, which critically contributes to the molecular coupling between ligand binding and channel opening. In this study, we report the identification and characterization of five previously unidentified prokaryote homologs with high sequence similarity (24-32%) to eukaryote HCN and CNG channels and that contain a C-linker region. Biochemical characterization shows that two homologs, termed AmaK and SthK, can be expressed and purified as detergent-solubilized protein from Escherichia coli membranes. Expression of SthK channels in Xenopus laevis oocytes and functional characterization using the patch-clamp technique revealed that the channels are gated by cAMP, but not cGMP, are highly selective for K(+) ions over Na(+) ions, generate a large unitary conductance, and are only weakly voltage dependent. These properties resemble essential properties of various eukaryote HCN or CNG channels. Our results contribute to an understanding of the evolutionary origin of cyclic nucleotide-modulated ion channels and pave the way for future structural and functional studies.

Keywords: cyclic AMP; cyclic GMP; electrophysiology; protein purification.

Fig. 1.

Sequence analysis of prokaryote cyclic nucleotide-modulated ion channels. (A) Sequence alignment of five prokaryote homologs with spHCN1, human HCN2, human CNGA1, human CNGB1, and MloK1. The cartoon represents the transmembrane topology of a single-channel subunit. In the S4 segment, positively charged residues are colored in red. Residues are colored in shades of blue by using an identity threshold of 50%. Sequence features are annotated according to the secondary structures features described in the X-ray crystal structure of the HCN2 C-linker region (A′–F′) and cyclic nucleotide binding domain (A–C) (6). (B) Cartoon representation of the HCN2 C-linker region and cyclic nucleotide binding domain (6). The left panel is a single subunit, and the right panel is a tetrameric assembly. cAMP molecules are shown as spheres (white, carbon; red, oxygen; blue, nitrogen; orange, phosphor). (C) Phylogenetic relationship of the homologs described in A. (D) Pairwise sequence identities of the homologs described in A.

Fig. 2.

Biochemical characterization of prokaryote cyclic nucleotide-modulated ion channels. (A) Cartoon presentation of the C-terminal GFP-fusion constructs used for detergent screening. (B) Whole-cell fluorescence obtained from E. coli cultures expressing AmaK-GFP in different strains. Fluorescence intensity was normalized relative to KcsA-GFP (29). (C) FSEC profile (28) of KcsA-GFP after solubilization with 2% (wt/vol) undecylmaltoside and gel filtration in running buffer containing 0.15% undecylmaltoside. (D) FSEC profile of AmaK-GFP after solubilization under the same conditions as in C. The gel scan shows in-gel green fluorescence of His-trap purified protein (lane b) loaded on a 4–15% Miniprotean TGX gel. Lane a contains Precision Plus All Blue standard from Bio-Rad. (E) FSEC profile of SthK-GFP after solubilization with 2% dodecylmaltoside and gel filtration in running buffer containing 0.05% dodecylmaltoside. The gel scan shows in-gel green fluorescence under the same conditions as in D except that an additional ladder (Fluorescent WesternC standard from Bio-Rad) was loaded in lane b. Lane c was loaded with His-trap purified protein.

Fig. 3.

Ligand selectivity in inside-out macropatches. (A) A voltage ramp ranging from −130 to +130 mV in 4.5 s was applied to an inside-out macropatch excised from a SthK-GFP–expressing Xenopus laevis oocyte. The black trace represents the current response after applying 500 µM cAMP at symmetrical potassium. There was no current response after removal of cAMP or applying 5 mM cGMP to the same patch, shown as gray and red traces, respectively. (B) The concentration–activation relationship for cAMP was obtained at symmetrical potassium. Means were calculated from 3 to 14 recordings. Error bars indicate SEM. The black curve is the best fit of the mean values using Eq. S1. When fitting single recordings, we obtained the following: EC₅₀ = 3.68 ± 0.55 µM (n = 10), H = 1.33 ± 0.08 (n = 10). (C) A voltage ramp under similar conditions as described in A was applied to an inside-out macropatch. The representative traces show current responses after applying 3 µM cAMP (black traces), 3 µM cAMP plus 5 mM cGMP (orange trace), or 5 mM cGMP (red trace). The ligands were washed in the same order as shown. There was no current response as long as cGMP was applied.

Fig. 4.

Functional characteristics of SthK in inside-out patches. (A) (Left) Current–voltage relationship as measured under four different ionic conditions. Bath solution and pipette solution contained either K⁺ or Na⁺ ions, respectively. The bath solution additionally contained a saturating cAMP concentration of 500 µM. Positive values indicate outward currents, and negative values indicate inward currents. Current amplitudes are normalized to the maximum value obtained at +100 mV at K_int/K_ex and K_int/Na_ex, respectively. Means were calculated from three to nine patches. Error bars indicate SEM. (Right) Representative currents (black traces: at 500 µM cAMP; gray traces: no cAMP) for each of the four ionic conditions. (B) Single-channel events were recorded at a saturating concentration of 500 µM cAMP at symmetrical potassium and +100 mV. (Left) Representative current–amplitude histogram for a three-channel patch. The main histogram was fitted by the sum of three Gaussian functions (for one closed and two open levels) (black line). The amplitude of the single-channel current was determined to be 8.1 pA (with a SD obtained from the Gaussian fit of 0.02 pA). The Inset shows the same histogram at higher magnification to visualize the third open level peak. This histogram was fitted by the sum of two Gaussian functions (black line). (Right) Shown are representative traces from the same recording as shown in the left panel. Closed and open levels are indicated as dashed (c) and dotted lines (o).