Ligand binding and activation properties of the purified bacterial cyclic nucleotide-gated channel SthK

Abstract

Cyclic nucleotide-modulated ion channels play several essential physiological roles. They are involved in signal transduction in photoreceptors and olfactory sensory neurons as well as pacemaking activity in the heart and brain. Investigations of the molecular mechanism of their actions, including structural and electrophysiological characterization, are restricted by the availability of stable, purified protein obtained from accessible systems. Here, we establish that SthK, a cyclic nucleotide-gated (CNG) channel from Spirochaeta thermophila, is an excellent model for investigating the gating of eukaryotic CNG channels at the molecular level. The channel has high sequence similarity with its eukaryotic counterparts and was previously reported to be activated by cyclic nucleotides in patch-clamp experiments with Xenopus laevis oocytes. We optimized protein expression and purification to obtain large quantities of pure, homogeneous, and active recombinant SthK protein from Escherichia coli A negative-stain electron microscopy (EM) single-particle analysis indicated that this channel is a promising candidate for structural studies with cryo-EM. Using radioactivity and fluorescence flux assays, as well as single-channel recordings in lipid bilayers, we show that the protein is partially activated by micromolar concentrations of cyclic adenosine monophosphate (cAMP) and that channel activity is increased by depolarization. Unlike previous studies, we find that cyclic guanosine monophosphate (cGMP) is also able to activate SthK, but with much lower efficiency than cAMP. The distinct sensitivities to different ligands resemble eukaryotic CNG and hyperpolarization-activated and cyclic nucleotide-modulated channels. Using a fluorescence binding assay, we show that cGMP and cAMP bind to SthK with similar apparent affinities, suggesting that the large difference in channel activation by cAMP or cGMP is caused by the efficacy with which each ligand promotes the conformational changes toward the open state. We conclude that the functional characteristics of SthK reported here will permit future studies to analyze ligand gating and discrimination in CNG channels.

Figure 1.

SthK topology and purification details. (A) Cartoon representation of the HCN1 channel structure (Lee and MacKinnon, 2017; Protein Data Bank accession no. 5U6O) as a model to highlight the domain architecture of the homologous SthK channel. One subunit of the channel is shown with the S1–S4 in blue, S5–S6 in purple, the C-linker in yellow, the CNBD in green, and the remaining three subunits in gray. The figure was prepared using PyMol (http://www.pymol.org). (B) Elution profile from size-exclusion chromatography (Superdex 200 10/300 GL) for SthK in the presence (solid line) and absence (dashed line) of cAMP; the inset shows an SDS-PAGE analysis of purified SthK (lane 1) and the results of cross-linking with 0.12% glutaraldehyde, confirming tetrameric assembly of SthK (lane 2). BenchMark Prestained Protein Ladder (Life Technologies) was used to estimate the molecular weight. (C) A representative negative-stain EM micrograph (bar, 141 nm) is shown to illustrate the quality of the final protein sample. A few representative particles used for 2-D classification are highlighted (small box) and enlarged in the inset (top, right; bar, 10 nm). (D) The resulting 2-D classes (box size, 248 Å) from negative-stain images are shown.

Figure 2.

Characterization of SthK channel activity. (A) Normalized ⁸⁶Rb⁺ flux through SthK (reconstituted in 5:3:2 DOPC:POPG:cardiolipin) in the presence of 200 µM cAMP (red squares), cGMP (blue circles), both cAMP and cGMP (black triangles), and without cyclic nucleotides (open triangles). Flux through empty liposomes (open circles) is shown as a reference. All flux values are normalized to the maximum uptake recorded in the presence of valinomycin. Bars represent means ± SEM for three separate experiments. (B) SthK channel activity represented by the maximum achieved ⁸⁶Rb⁺ uptake through SthK reconstituted in liposomes made from 3:1 POPE:POPG (PE:PG) or 5:3:2 DOPC:POPG:cardiolipin (PC:PG:CA), as indicated, in the presence of 200 µM cAMP. Bars represent means ± SEM for three separate experiments. (C) Representative single-channel recordings of SthK in 0.1 mM intracellular cAMP. (D) I-V relationship of SthK single-channel current amplitude at 100 µM cAMP. (E) Open probability of SthK at 100 mV as a function of the cAMP concentration. Fitting the data with Eq. 10 yields an apparent activation constant of EC₅₀ = 17 µM and a Hill coefficient of n_H = 3. The number of repeats is indicated beside each data point. (F) Open probability of SthK in the presence of 500 µM cAMP as function of the membrane potential. The line indicates the fit according to Eq. 13, leading to z = 0.8 and V_half = 87 mV, Po_max = 0.65, and Po_min= 0.05. Symbols and error bars in D–F represent means ± SEM from at least three separate bilayers.

Figure 3.

Kinetics of SthK activation by cAMP and inhibition by cGMP. (A) Fluorescence decay traces as a measure of channel activity in the stopped-flow Tl⁺ flux assay after incubating the SthK liposomes with 200 µM cAMP for 15 ms (blue), 100 ms (red), and 5 s (green). 0 µM cAMP is in black. The corresponding flux rates (Eq. 9) as a function of activation time are shown in D. (B) Fluorescence decay traces from SthK liposomes incubated for 2.5 s with cAMP over a range of concentrations (0, 25, 50, 100, and 400 µM from top to bottom). The corresponding flux rates as a function of the cAMP concentration are shown in E. (C) Fluorescence decay traces from SthK liposomes incubated with 200 µM cAMP and increasing concentrations of cGMP for 2.5 s (0 µM [green], 25 µM [blue], 100 µM [turquoise], 500 µM [yellow], and 1,000 µM [red]). No cyclic nucleotides (black). The corresponding flux rates as a function of the cGMP concentration are shown in F. (D) The activation time course is best fitted with a sum of two exponentials (line through symbols) with time constants of τ₁ ≈ 0.05 s (amplitude 0.45) and τ₂ ≈ 1.04 s (amplitude 0.55). (E) A fit with Eq. 10 yields an EC₅₀ of ∼100 μM (line through symbols) and a Hill coefficient of 2.2 ± 0.3. (F) The apparent inhibition constant was determined using Eq. 11 (line through symbols) and yields an IC₅₀ of 40 µM, from which the K_i of 6 µM was calculated (Eq. 12; Table 1). (D–F) Rates were normalized to the maximum value, and symbols and error bars represent means ± SEM for three experiments.

Figure 4.

cGMP inhibits cAMP-dependent channel activity. (A) Representative single-channel traces of SthK at 100 mV in the presence of 100 µM cAMP (top trace). Addition of 1 mM cGMP (middle trace) significantly, and reversibly, reduces channel activity. Application of only cGMP (bottom trace) leads to low but nonzero channel activity. (B) Inhibition of preactivated SthK (by 100 µM cAMP) with increasing concentrations of cGMP. All data analyzed are from recordings at 100 mV. Data are fitted using Eq. 11, giving an IC₅₀ of 16 µM. Symbols and error bars represent means ± SEM for three experiments.

Figure 5.

Interaction of cAMP and cGMP with SthK. (A) Titration of 0.1 µM fcAMP (closed circles) or fcGMP (open circles) with SthK in A8-35; lines represent fits according to Eq. 4, giving K_d values of 0.4 µM and 2 µM, respectively. The data presented are from a single measurement; averaged values from three separate titrations are listed in Table 1. (B) Representative titration of the SthK · fcAMP complex with increasing concentrations of cAMP (closed circles) or cGMP (open circles). The K_d values for the unlabeled cyclic nucleotides obtained from these titrations according to Eq. 6 are 50 ± 8 µM for cAMP and 19 ± 4 µM for cGMP. Averaged values from three competition experiments are given in Table 1. (C and D) Overlay of the CNBDs of HCN2 (apo [blue], 5JON [Goldschen-Ohm et al., 2016]) and SthK in complex with cAMP (orange, 4D7T [Kesters et al., 2015]) in C, and the same HCN2 and SthK in complex with cGMP (cyan, 4D7S [Kesters et al., 2015]) in D. The figures were prepared using PyMol (http://www.pymol.org). The ligands cAMP (C) and cGMP (D) are shown in stick representation in CPK colors. Only E421 is shown to indicate an interaction with cAMP (C) but not with cGMP (D).