Model prokaryotic potassium channels: Predictive power and relevance to eukaryotic channels

Ion channels are essential to biological systems, orchestrating numerous cellular functions by regulating the flow of ions across membranes (Hodgkin and Huxley, 1952b). These transmembrane proteins act as molecular switches that respond to stimuli, including voltage changes, ligand binding, temperature, and mechanical stress. In excitable cells, ion channels are fundamental to action potentials, and the precisely coordinated activity of voltage-gated sodium and potassium (K⁺) channels facilitates the propagation of nerve impulses over long distances (Hille et al., 1999; Hille, 2001; Armstrong, 2007). Similarly, ion channels play a crucial role in sensory transduction, converting external signals, such as light in retinal photoreceptors or mechanical pressure in cochlear hair cells, into electrical responses that the nervous system can interpret. In the context of metabolism, ion channels contribute to cellular energy homeostasis. Given their many physiological roles, ion channel dysfunction is linked to a range of pathologies, including neurological disorders such as epilepsy and migraine, cardiac arrhythmias, cystic fibrosis, and cancer, underscoring the importance of elucidating ion channel mechanisms from both physiological and medicinal perspectives (Ashcroft, 1999; Ashcroft, 2000; Ashcroft, 2006).

K⁺ channels, the largest family of ion channels, are generally modular in structure (MacKinnon, 1991b; MacKinnon, 2003) (Fig. 1). Understanding the workings of K⁺ channels is thus similar to assembling a puzzle piece by piece, by examining one module at a time. Eukaryotic K⁺ channels are typically multi-subunit complexes, posing considerable experimental challenges. Many are difficult to express in recombinant systems, and their regulation is generally polymodal, involving crosstalk among stimuli such as voltage sensing, ligand binding, posttranslational modifications, and interactions with accessory proteins, complicating biophysical analysis and detailed mechanistic studies. Consequently, there is a need for systems that can replicate the core biophysical properties of ion channels while circumventing the intricacies of eukaryotic systems.

Homologous bacterial ion channels have served and continue to serve as valuable biophysical model systems for their eukaryotic counterparts due to their relative structural simplicity and ample sample availability (MacKinnon et al., 1998). Prokaryotic channels typically exhibit homomeric, compact architectures that lack auxiliary domains, regulatory protein–protein interaction motifs, and obligatory posttranslational modifications. This reduced complexity facilitates the study of isolated functional elements and gating mechanisms. Bacterial channels can be expressed in Escherichia coli, yielding milligram quantities of functional protein, in stark contrast to eukaryotic channels, which often require specialized eukaryotic expression systems. The high yield of pure, homogeneous prokaryotic channel proteins is particularly beneficial for resource-intensive biophysical techniques and in vitro functional studies. They can be readily reconstituted into artificial lipid bilayers or liposomes, providing a controlled environment for measuring functional properties (Schmidpeter et al., 2020b).

Arguably, the most transformative contribution of bacterial ion channels lies in structural biology. Since the groundbreaking crystal structure of KcsA (Doyle et al., 1998), prokaryotic channels have provided a wealth of high-resolution structural data (first obtained by x-ray crystallography and later by cryogenic electron microscopy [cryo-EM]) on key features conserved in eukaryotic K⁺ channels. While cryo-EM has, in the past decade, rapidly yielded vast amounts of structural information on eukaryotic ion channels, a detailed correlation of these structures with functional data under the same experimental conditions is often limited by sample constraints. Collectively, the simplicity, high yield, functional tractability, and structural accessibility of bacterial ion channels reinforce their enduring value as model systems.

Many prokaryotic systems, including voltage-gated channels, pentameric ligand-gated ion channels, and sodium channels, have been identified and studied over the years. In addition to their value as biophysical model systems, bacterial ion channels also offer critical insights into microbial physiology, including ion homeostasis and adaptation to environmental stress (Compton and Mindell, 2010; Stautz et al., 2021). These functions highlight that prokaryotic channels are not merely simplified analogs but biologically relevant systems that provide insight into ion transport across diverse contexts.

Here, we discuss select prokaryotic K⁺ channels to evaluate their value and limitations in predicting functional and regulatory details of their eukaryotic homologs. Our outline will follow the modular architecture of K⁺ channels (Fig. 1 A) and briefly address the pore domain, which is conserved across all K⁺ channels. The next, more complex structure includes cytosolic ligand-binding domains that modulate the pore domain. Finally, additional transmembrane domains enable further modes of regulation. This increasing structural complexity is reflected in the prokaryotic channels KcsA, MthK, and SthK (Fig. 1, A and B). Work on these bacterial channels has led to detailed mechanisms of gating and regulation that, in large part, apply to their eukaryotic counterparts. This Review highlights the crucial role of prokaryotic K⁺ channels in bridging the gap between fundamental biophysical insights and the intricate regulatory landscapes of eukaryotic channels.

The first cloned K⁺ channel, Shaker, was named after a phenotype observed in Drosophila melanogaster (Jan et al., 1977; Salkoff and Wyman, 1981; Kamb et al., 1987; Tempel et al., 1987). Mutations in this gene caused ether-induced leg-shaking behavior, mimicking the effects of K⁺ channel blockers (Jan et al., 1977). Cloning of the Shaker gene rapidly generated extensive knowledge about K⁺ channel functionality (Kamb et al., 1987; Tempel et al., 1987). Collaborative efforts from many research groups facilitated the discovery of fundamental principles governing K⁺ channels, including rapid inactivation, preference for K⁺ over other monovalent cations, and the identification of the selectivity filter (Hoshi et al., 1990; Heginbotham and MacKinnon, 1993; Heginbotham et al., 1994; Doyle et al., 1998; MacKinnon, 2003). As additional sequence data became available for more K⁺ channels, the highly conserved selectivity filter motif—TVGYG—emerged as the defining sequence for these channels (MacKinnon et al., 1990; Yool and Schwarz, 1991; Heginbotham et al., 1994; Doyle et al., 1998; Shealy et al., 2003). Additionally, it was established that K⁺ channels are translated into smaller proteins compared with sodium channels and must form tetrameric complexes to function (Fig. 2, A–E) (MacKinnon, 1991a).

However, many critical aspects of K⁺ channel function remained hypothetical in the absence of atomic-resolution structural information. KcsA, a bacterial channel from Streptomyces lividans, was the first K⁺ channel to enable solving a high-resolution crystal structure of these proteins, one of the most influential results in ion channel biophysics (MacKinnon and Doyle, 1997; Doyle et al., 1998; LeMasurier et al., 2001; Zhou et al., 2001). It revealed the permeation pathway and the core architecture of all K⁺ channels: two transmembrane helices per monomer, arranged around an ion conduction pathway in fourfold symmetry (Fig. 2, A–C). At the intracellular side, these helices come together in a bundle crossing, a constriction that gates many K⁺ channels. KcsA channels open in response to acidic pH (Cuello et al., 1998), where protonation engages a network of ionizable residues near the bundle-crossing gate (Thompson et al., 2008; Cuello et al., 2010a; Posson et al., 2013b). However, different stimuli, such as voltage, Ca²⁺, and various other ligands, can, through distinct mechanisms, open the same bundle-crossing gate (Yellen, 1998). The pore-lining helices create a water-filled inner vestibule, which is a continuation of the cytosol, allowing ions to enter this space without an energetic penalty and effectively minimizing their travel distance across the hydrophobic membrane core. This is further aided by a favorable electric field generated by the dipoles of the four pore helices, which point toward the cavity center (Roux and MacKinnon, 1999). Between the inner vestibule and the outer mouth of K⁺ channels, the selectivity filter provides the central, most conserved structural element of all K⁺ channels. Four consecutive backbone carbonyl groups from each monomer, provided by residues of the signature sequence of K⁺ channels (GYG), point toward the ion conduction pathway, forming four octahedral cages arranged in series, each an almost perfect substitute for the hydration shell of a K⁺ (Fig. 2, D and E) (Doyle et al., 1998; Zhou et al., 2001). This arrangement and the precise carbonyl coordination are responsible for the high selectivity for K⁺ over Na⁺ (Morais-Cabral et al., 2001; Noskov et al., 2004; Thompson et al., 2009). This architecture is conserved among all K⁺ channels with thus far elucidated structures (Fig. 2 F), and to this day, the structure of KcsA serves as an essential reference for all K⁺ channels.

Beyond being a defining element for K⁺ channels, the selectivity filter domain was also proposed to be involved in gating mechanisms, such as C-type inactivation (Choi et al., 1991; Hoshi et al., 1991; Perozo et al., 1999; Zhou et al., 2001; Cordero-Morales et al., 2006; McCoy and Nimigean, 2012). In KcsA, C-type inactivation was proposed to be a structural deformation (narrowing or “collapse”) of the selectivity filter that renders the channel nonconductive (Ogielska and Aldrich, 1999; Zhou et al., 2001; Cuello et al., 2010b; Cheng et al., 2011; Bhate and McDermott, 2012; Liu et al., 2015). Computational studies and mutagenesis experiments support models in which subtle structural changes in the filter region disrupt K⁺ coordination (Bernèche and Roux, 2005; Cheng et al., 2011; Matulef et al., 2013; Boiteux et al., 2020; Kopec et al., 2023; Lau et al., 2024). Recent structural data have proposed a different type of deformation: a slight widening rather than narrowing of the filter (Domene et al., 2008; Devaraneni et al., 2013; Hoshi and Armstrong, 2013; Kopec et al., 2019; Reddi et al., 2022; Tan et al., 2022; Stix et al., 2023; Wu et al., 2025). Regardless of the minute details, any deformations in the carbonyl cage geometry of the filter will disrupt K⁺ coordination and thus, presumably, lead to C-type inactivation (Hoshi and Armstrong, 2013). Therefore, the original principles of C-type inactivation based on KcsA structures, which describe a set of conformational changes centered on the selectivity filter that disrupt K⁺ coordination, were shown to be conserved in the eukaryotic counterparts.

Even today, 30 years after its discovery, KcsA remains a cornerstone in the ongoing exploration of ion channels, contributing to many scientific discoveries related to ion permeation and K⁺ channel regulation. Its minimalistic structure, ease of manipulation, and the wealth of available data make it an invaluable system for studies using techniques such as two-dimensional infrared spectroscopy, diffracted x-ray tracking methods with gold nanocrystals, solid-state NMR, and all-atom molecular dynamics simulations (Shrivastava and Sansom, 2000; Shimizu et al., 2008; Pan et al., 2011; Sun et al., 2020; Ryan et al., 2023). These insights continue to inform our understanding of K⁺ channel function. In summary, KcsA is more than just a structural blueprint. It played a pivotal role in elucidating the intricate physiology of ion transport across cellular membranes, leading to significant advances in our understanding of eukaryotic K⁺ channels.

In addition to the pore domain, many K⁺ channels have regions outside the membrane that can sense regulatory ligands, such as Ca²⁺, which is essential for cellular signaling (Fig. 1). In eukaryotes, large-conductance Ca²⁺-gated K⁺ (BK) channels are crucial for various physiological functions, including control of cellular excitability, muscle contraction, and neurotransmitter release. Dysfunction in BK channels is associated with conditions such as hypertension, diabetes, and certain neurological disorders (Sancho and Kyle, 2021; Echeverría et al., 2024). Beyond the conserved pore domain, these channels have cytosolic Regulator of Conduction of K⁺ (RCK) domains that bind Ca²⁺ ions, triggering conformational changes that open the pore (Jiang et al., 2001; Giraldez and Rothberg, 2017). Additionally, shifts in membrane potential and interactions with regulatory subunits affect BK channel activity (Dworetzky et al., 1994; Gonzalez-Perez and Lingle, 2019), adding additional layers of complexity that make detailed biophysical studies of these proteins more difficult.

The bacterial channel MthK from Methanobacterium thermoautotrophicum has been recognized as a biophysical model system for studying Ca²⁺-dependent regulation of BK channels (Shin et al., 2001; Dong et al., 2005; Parfenova et al., 2006; Ye et al., 2006; Ye et al., 2010; Zadek and Nimigean, 2006; Li et al., 2007; Pau et al., 2010; Pau et al., 2011; Thomson and Rothberg, 2010; Shi et al., 2011; Posson et al., 2013a; Posson et al., 2015; Thomson et al., 2014; Boiteux et al., 2020; Fan et al., 2024) since the first report of its x-ray crystallographic structure (Jiang et al., 2002a; Jiang et al., 2002b). Structurally, MthK resembles a simplified BK channel, consisting of a K⁺ channel core (i.e., the pore domain) and cytosolic RCK domains (Figs. 1, 3 A). X-ray crystallographic studies of Ca²⁺-bound MthK provided insights into ligand-dependent gating by comparing it with the closed state of KcsA and by modeling how intracellular domains may connect ligand binding to pore opening (Fig. 3, B and C). Upon Ca²⁺ binding, conformational changes in the cytosolic domains result in an outward rotation of the pore-lining helices, leading to a widening of the ion conduction pathway at the intracellular entryway, called the bundle-crossing gate (Fig. 3, B and C). This was later confirmed by obtaining the Ca²⁺-free cryo-EM structure of MthK (Fan et al., 2020), which indeed showed a closed bundle crossing similar to KcsA, further confirming the conservation of the pore and gate domains between channels (Fig. 3, A–C). The linker region between the TM2 pore-lining helices and the RCK domains could not be resolved in the initial crystallographic studies, but cryo-EM of MthK in lipid nanodiscs provided further insights into Ca²⁺-dependent gating movements. The conformational changes associated with Ca²⁺ binding (Jiang et al., 2002b; Ye et al., 2006) lead to unfolding of the linker helices connecting the RCK and pore domains, allowing increased mobility of the gating ring with reduced orientational restriction (Fig. 3 D) (Jiang et al., 2002a; Fan et al., 2020). The availability of multiple structural states of MthK, along with functional data obtained under identical conditions, enabled mathematical modeling of general gating energetics in these channels (Scheuring, 2023).

Bundle-crossing gating, first structurally shown in KcsA and MthK, is also how many eukaryotic K⁺ channels were originally proposed to gate and, more recently, shown to gate using cryo-EM structural studies (Tao et al., 2009; Hansen et al., 2011; Meng et al., 2016; Zhao and MacKinnon, 2021). However, the bundle crossing may not function as a universal gate for all K⁺ channels. For example, gating in BK channels, the eukaryotic homologs of MthK, was originally proposed to occur at the selectivity filter, based on functional assays implicating conformational changes in the selectivity filter in regulating ion conduction (Li and Aldrich, 2004; Piskorowski and Aldrich, 2006; Wilkens and Aldrich, 2006; Tang et al., 2009; Thompson and Begenisich, 2012). Cryo-EM structures of Ca²⁺-free BK channels display a large opening at the intracellular gate, where bundle crossing normally occurs (Fig. 3 C), large enough for K⁺ to flow through, apparently supporting the selectivity filter gate hypothesis. While these observations are often interpreted in terms of distinct gating mechanisms, selectivity filter– and bundle crossing–based gating are not necessarily mutually exclusive and may instead operate in a coupled, state-dependent manner (Kopec et al., 2019). Alternative hypotheses, such as hydrophobic and lipid gating, were also proposed as mechanisms for BK channel closure (Jia et al., 2018; Coronel et al., 2025; Mironenko et al., 2025, Preprint).

Another example where the bundle crossing does not necessarily act as a gate is the two-pore-domain K⁺ channels (K2P), which are mainly gated at the selectivity filter, similar to BK channels (Bagriantsev et al., 2011; Piechotta et al., 2011; Miller and Long, 2012; Dong et al., 2015; Lolicato et al., 2020; Neelsen et al., 2024). While the pore-lining helices can display movements that narrow or expand the bundle crossing, it is believed to function as a gate only in specific members of the K2P family and only for certain stimuli, such as voltage (Lolicato et al., 2014; McClenaghan et al., 2016; Zhuo et al., 2016). This dual modulation at the filter and the bundle crossing can allow effective integration of various stimuli (Bagriantsev et al., 2011; Kopec et al., 2019; Schmidpeter et al., 2023; Neelsen et al., 2024). Other K2P channels are thought to be regulated by an X-gate located just below the selectivity filter, in addition to the filter gating (Rödström et al., 2020; Hall et al., 2025). Other examples of alternative access into the pore are provided by MthK and BK channels, where state-dependent, membrane-facing fenestrations have been observed that can provide access to the pore when the bundle crossing is closed (Hille, 1977; Jorgensen et al., 2016; Fan et al., 2020; Fan et al., 2024; Contreras et al., 2025; Coronel et al., 2025; Mironenko et al., 2025, Preprint). Functional, structural, and computational studies suggest that even partially charged molecules, such as tetrapentylammonium, can enter the pore through these fenestrations via the lipid bilayer (Fan et al., 2024). In eukaryotic BK channels, these fenestrations may allow lipids to penetrate close to the pore to assist with hydrophobic gating. In summary, despite observations that the bundle crossing is not a universal gate in all K⁺ channels, an open bundle crossing is essential for unrestricted ion flow into the inner vestibule across all K⁺ channels.

In addition to opening and closing, ion channels can also inactivate, a process in which permeation is stopped while the activation stimulus remains present, which is necessary for precisely timed electrical signaling. Initially discovered in voltage-gated sodium channels (Hodgkin and Huxley, 1952a; Armstrong and Bezanilla, 1977; Bezanilla and Armstrong, 1977), inactivation was also quickly discovered in Shaker and many other K⁺ channels (Armstrong et al., 1973; Vassilev et al., 1988; Stuhmer et al., 1989; Hoshi et al., 1990; Hoshi et al., 1991; Choi et al., 1991; Ogielska and Aldrich, 1999; Cuello et al., 2010b; Pan et al., 2011). Elegant biochemical and electrophysiological experiments led to the proposal that in Shaker channels, this inactivation was caused by the N terminus of the channel acting to obstruct the pore, akin to a ball attached to the channel protein via a chain, rendering the channel nonconductive (Hoshi et al., 1990; Zagotta et al., 1990; Demo and Yellen, 1991). This mechanism was thus referred to as N-type or ball-and-chain inactivation. Many different K⁺ channels were subsequently shown to inactivate via this mechanism, by using either their own N termini or the N termini of accessory subunits that co-assemble with the channel (Wallner et al., 1999; Xia et al., 1999). Although compelling functional evidence for this mechanism was available from electrophysiological measurements, structurally resolving this critical conformation has remained elusive until its discovery in MthK channels (Fan et al., 2020), despite the growing body of high-resolution cryo-EM structures of many K⁺ channels that have been shown to inactivate via this mechanism.

The detailed structural analysis of the bacterial channel MthK using cryo-EM has filled this crucial gap in the conformational landscape of K⁺ channels, marking another milestone in K⁺ channel research enabled by a prokaryotic model system (Fan et al., 2020). The first 17 residues at the N terminus of the MthK channel form a flexible peptide that, following Ca²⁺-induced opening of MthK, swings into the channel pore to plug it, thus effectively blocking ion flow (Fig. 3 D). The first half of the N terminus is mainly hydrophobic and interacts strongly with the hydrophobic pore cavity, while the second half is mostly charged and presumably interacts with critical residues at the inner mouth of the pore, as was previously shown by decades of functional experiments on eukaryotic K⁺ channels (Hoshi et al., 1990; Solaro and Lingle, 1992; Antz et al., 1997; Wissmann et al., 1999; Sukomon et al., 2023).

The structural correlates of N-type inactivation obtained for the bacterial channel MthK brought to light a long-standing conceptual hypothesis relevant to many eukaryotic K⁺ channels, but it took another 5 years before ball-and-chain inactivation was structurally shown in a eukaryotic channel (Agarwal et al., 2025), again highlighting the difficulties in handling eukaryotic channels (Hoshi et al., 1990; Zagotta et al., 1990; Demo and Yellen, 1991; Wallner et al., 1999; Fan et al., 2020). The human BK channel in complex with an accessory, inactivating β-subunit was captured in a ball-and-chain inactivated conformation (Fig. 3 A) (Agarwal et al., 2025). Similar to MthK, a peptide at the cytosolic N terminus of the β2-subunit inserts into the open BK channel pore to plug the permeation pathway. Here, the connection (the “chain”) between the inactivating peptide and the transmembrane region is much longer than in MthK and was not resolved, suggesting it is disordered (Fig. 3, D and E). Recently, the ball-and-chain inactivated Shaker channel structure was also resolved with cryo-EM (Tan et al., 2025), with the N terminus expectedly lodged in the pore, as observed in the previously determined structures (Fan et al., 2020; Agarwal et al., 2025). More, similarly inactivated structures will likely be discovered in the future.

Bacterial homologs have provided extensive information on the structure and function of K⁺ channels applicable to nearly all K⁺ channels, as summarized above with select examples. However, as the complexity and functional specialization of K⁺ channels increase (Fig. 1), additional model systems are necessary to capture the characteristics of individual channel families. One example is the family of cyclic nucleotide–modulated channels, which includes cyclic nucleotide–gated (CNG) channels and the closely related hyperpolarization-activated cyclic nucleotide–modulated (HCN) channels, critical for sensory transduction and pacemaking, respectively (Kaupp and Seifert, 2001; Kaupp and Seifert, 2002; Craven and Zagotta, 2006; Biel et al., 2009).

All members of the cyclic nucleotide–modulated channel family share a common structure (Fig. 4). In addition to the pore domain, each monomer includes a voltage-sensing domain (VSD), a cytosolic C-linker, and a cytosolic cyclic nucleotide–binding domain (CNBD) (Fig. 4). Unlike most voltage-gated K⁺ channels, the VSDs in CNG and HCN channels are non–domain-swapped, meaning that conformational changes within a VSD influence the pore domain of the same subunit (Fig. 4 B) (Biel et al., 1993; Ludwig et al., 1999; Sunderman and Zagotta, 1999; Santoro et al., 2000; James et al., 2017; Lee and MacKinnon, 2017; Zheng et al., 2020). Conversely, a domain swap occurs at the level of the C-linker/CNBDs, leading to a characteristic “elbow-on-shoulder” arrangement (Fig. 4 C) (Zagotta et al., 2003; James and Zagotta, 2018). CNG channels are directly activated by cAMP and/or cGMP, while in HCN channels, cAMP binding biases voltage-dependent activation, facilitating channel opening. These findings, along with other functional details of CNG and HCN channels, are available from electrophysiological recordings on cells heterologously expressing these channels (Moroni et al., 2000; Santoro et al., 2000; Wainger et al., 2001; Rothberg et al., 2002; Wang et al., 2002; Xu et al., 2010).

Over the past 10 years, cryo-EM structures have been reported for both prokaryotic and eukaryotic CNG and HCN channels, including heteromeric CNG channels derived from native tissue (James et al., 2017; Lee and MacKinnon, 2017; Lee and MacKinnon, 2019; Rheinberger et al., 2018; Zheng et al., 2020; Zheng et al., 2022; Saponaro et al., 2021; Barret et al., 2022; Xue et al., 2022). The structures of these channels are strikingly similar (Fig. 4 A), yet functional data from purified channels under comparable conditions are often lacking, due to either insufficient sample amounts, a lack of suitable functional assays, or unknown cofactor requirements for activity (such as the “correct” lipid composition). Due to the structural and functional conservation with its eukaryotic counterparts, the prokaryotic SthK channel from Spirochaeta thermophila has emerged as a valuable biophysical model for cyclic nucleotide–modulated channels (Fig. 4, A and D). Its initial discovery was accompanied by electrophysiological current measurements from Xenopus laevis oocytes expressing the SthK gene, confirming its function as a CNG channel. Subsequent optimization of the expression construct and conditions enabled the purification of functional SthK channels for biophysical studies (Brams et al., 2014; Kesters et al., 2015; Schmidpeter et al., 2018; Morgan et al., 2019). SthK is directly activated by cAMP binding, and its open probability increases with depolarization, similar to eukaryotic CNG channels, validating it as a functional model (Goulding et al., 1994; Varnum et al., 1995; Marchesi et al., 2012; James and Zagotta, 2018; Schmidpeter et al., 2018; Morgan et al., 2019). Furthermore, despite similar binding affinities, cAMP and cGMP differentially regulate SthK activity, as observed in various eukaryotic CNG channels (Goulding et al., 1994; Varnum et al., 1995; Kaupp and Seifert, 2002; Brams et al., 2014; Schmidpeter et al., 2018; Schmidpeter and Nimigean, 2018; Morgan et al., 2019; Pan et al., 2023). The availability of pure, functional SthK has also enabled detailed studies of conformational rearrangements, structural dynamics, and allosteric coupling in CNG channels using double electron–electron resonance spectroscopy, transition metal FRET, single-molecule FRET, atomic force microscopy, and electrochemical impedance spectroscopy (Marchesi et al., 2018; Evans et al., 2020; Gao et al., 2022; Andersson et al., 2023; Pan et al., 2023; Eggan et al., 2024; Eggan et al., 2025). The wealth of data generated by this versatile set of tools has also led to the use of SthK for optogenetic manipulations of cyclic nucleotide signaling (Bernal Sierra et al., 2018; Henss et al., 2022; Spreen et al., 2025).

Structural data for SthK are available from x-ray crystallographic studies of the isolated C-linker/CNBDs and from cryo-EM data of full-length SthK (Kesters et al., 2015; Rheinberger et al., 2018). The structure of SthK closely resembles that of eukaryotic CNG and HCN channels (Fig. 4, A and D) (Lee and MacKinnon, 2017; Zheng et al., 2020). It includes all the conserved elements: a VSD (S1–S4), a pore domain (S5–S6), and a cytosolic CNBD connected to the pore via a C-linker. In combination with functional characterization, this conserved structure makes the bacterial protein a powerful tool for studying the gating of CNG and HCN channels under defined conditions.

The isolated C-linker/CNBD of SthK exhibited cAMP-dependent conformational changes similar to those proposed for CNG and HCN channels. In contrast, the full-length SthK channel initially yielded only closed-state structures, even in the presence of saturating cAMP, as expected from the low open probability of the channel at 0 mV due to the low efficacy of cyclic nucleotides to open SthK (Fig. 5 A) (Kesters et al., 2015; Rheinberger et al., 2018; Schmidpeter et al., 2018). Nevertheless, this structure represents a crucial step in the gating pathway: the agonist-bound resting state, virtually identical to the apo state. This state has not yet been observed in eukaryotic CNG channels, likely because cyclic nucleotides are highly efficacious at opening these channels.

Mutations designed to increase cAMP efficiency by mimicking the depolarized state of the voltage sensor led to the identification of not only the open state but also several on-path gating intermediates for SthK (Gao et al., 2022). One mutant, Y26F, disrupts the coordination of Arg111 in the voltage sensor, and the other, R120A, eliminates one of the putative voltage-sensing gating charges. Both residues presumably contribute to voltage sensing in SthK (Fig. 5 B). Detailed inspection of the multiple intermediate and open structures revealed that the increased activity of both mutant channels stems from the voltage sensor adopting a depolarized-like state as a consequence of the mutations, thereby releasing its inhibition of the pore (Gao et al., 2022). In WT SthK, membrane depolarization was thus proposed to release this voltage-sensor inhibition.

The conformational changes leading to SthK channel opening in response to cAMP are identical with those observed in eukaryotic CNG channels (Sunderman and Zagotta, 1999; Kaupp and Seifert, 2002; Zagotta et al., 2003; James and Zagotta, 2018; Zheng et al., 2020). The C-helix swings upward to interact with the bound cAMP in the binding pocket, and the entire CNBD moves vertically closer to the membrane and radially outward. The entire cytosolic domains, including the CNBD and C-linker, rotate counterclockwise and pull on the ends of the S6 helices to open the bundle crossing and dilate the pore (Fig. 5, C–E). These conformational changes are almost identical to those observed in eukaryotic CNG channels (Fig. 5, F–H). For SthK in particular, the S6 helices can only move outward to open the pore if the S4 voltage-sensing helix kinks outward to create space. This is evident in the intermediate SthK structures, which exhibit varying degrees of S4 displacement, resulting in different levels of pore opening (Fig. 5 E). These rather small S4 displacements are, perhaps surprisingly, sufficient to allow the S6 helices enough space to expand and open the pore, unlike the large VSD movements required to actively open the pore in voltage-gated channels (Glauner et al., 1999; Long et al., 2007; Mandala and MacKinnon, 2022). Nevertheless, equally small VSD movements have been observed in KAT1, a Kv channel from Arabidopsis thaliana that shares a similar domain architecture with CNG and HCN channels, as well as in BK channels (Clark et al., 2020; Contreras et al., 2025). Similar gating intermediates were later identified in heteromeric human cone photoreceptor CNGA3/CNGB3 and CNGA1 channels, suggesting a comparable conformational trajectory from closed to open channels as observed in SthK (Gao et al., 2022; Hu et al., 2023; Park and Nimigean, 2025, Preprint). For most examples (KAT1, BK, and CNGA3/CNGB3), functional data for the purified channels under the same conditions as the structural experiments remain elusive. In light of this, the conceptual data generated using SthK, which allows functional data, state distributions from single-molecule FRET studies, and cryo-EM structures to be directly compared, assume a central role in describing conformational transitions in CNG channels.

In addition to cyclic nucleotides and voltage, the activity of CNG and HCN channels is also modulated by the lipid environment. The signaling lipids phosphatidic acid (PA) and phosphatidylinositol 4,5-bisphosphate (PIP2) enhance HCN channels, whereas PIP2 inhibits CNG channels (Womack et al., 2000; Pian et al., 2006; Pian et al., 2007; Zolles et al., 2006; Fogle et al., 2007). However, the cellular membrane composition can be altered only qualitatively and only for select lipids in the cell-based electrophysiological assays available for these channels. One advantage of bacterial model channels, including SthK, over their eukaryotic counterparts is that pure channel proteins can be obtained in large quantities and used in various functional and biophysical assays by incorporating them into liposomes of defined lipid composition. This versatility has enabled detailed mechanistic studies of lipid modulation of SthK at the molecular level (Schmidpeter et al., 2022; Thon et al., 2024; Newton et al., 2025).

Stopped-flow flux assays and single-channel recordings showed that SthK activity is enhanced in the presence of negatively charged lipids such as phosphatidylglycerol, cardiolipin, or PA (Schmidpeter et al., 2022). In addition, mass spectrometry (MS)–based lipidomics and native MS showed that anionic lipids copurified with SthK from the bacterial membrane. Together, this indicates that lipids directly bind to and modulate SthK function. Further, cryo-EM studies revealed numerous lipid-like densities surrounding SthK (Rheinberger et al., 2018; Gao et al., 2022; Schmidpeter et al., 2022), including one lipid bound at a gating-sensitive location. The headgroup of this lipid is positioned near an inter-subunit salt bridge between the S5 and S6 helices of adjacent subunits, which can only form in the closed state and must break for the channel to open (Fig. 6, A and B). With anionic lipids bound at this location, the negatively charged headgroups destabilize this salt bridge and, consequently, the closed state, facilitating channel opening (Fig. 6) (Schmidpeter et al., 2022).

This inter-subunit salt bridge, which mediates the enhancement of SthK activity by anionic lipids, is strictly conserved in HCN channels (Fig. 6, C and D) but absent in CNG channels, allowing predictions about lipid modulation of eukaryotic pacemaker channels (Decher et al., 2004; Schmidpeter et al., 2022). However, neither stopped-flow assays nor single-channel recordings are feasible for purified HCN channels at this point due to their need for hyperpolarizing potentials for activation and small single-channel conductance. In this regard, the wealth of detailed information generated for SthK enabled the design of targeted experiments on eukaryotic HCN channels using available techniques to test this hypothesis. Two-electrode voltage-clamp recordings on Xenopus laevis oocytes expressing hyperpolarization-activated cyclic nucleotide–modulated channel 2 (HCN2), combined with mutational studies and pharmacological alterations in membrane composition, suggested that a mechanism similar to that identified for SthK also regulates eukaryotic HCN channels (Schmidpeter et al., 2022). Additionally, these experiments revealed that destabilizing the inter-subunit salt bridge at the bundle-crossing gate is possible only with PA, not with PIP2. This provides another example of how bacterial model channels can inform experimental design and aid in interpreting results for more complex eukaryotic homologs.

Results on SthK and HCN2 also enable hypothesis generation about lipid modulation of CNG channels. SthK D226N, in which the inter-subunit salt bridge is permanently abolished, was designed based on the sequence of CNG channels (Fig. 6 C). Quantitative lipidomics of this mutant SthK channel showed that breaking the salt bridge is sufficient to eliminate the preference of SthK for anionic lipids (Schmidpeter et al., 2022). This supports the hypothesis that CNG channels exhibit a distinct lipid profile and likely differ in lipid dependence from HCN channels.

Recently, it was found that lipids also contribute to temperature sensing in SthK (Li and Nimigean, 2026). SthK is robustly activated by cooling, but only in the presence of amine-containing lipids, through destabilization of the same inter-subunit salt bridge that modulates channel activity and long-range allosteric regulation of cAMP potency (Schmidpeter et al., 2022; Newton et al., 2025). This structural feature is conserved in HCN channels (Decher et al., 2004; Schmidpeter et al., 2022), indicating that related channels may harbor latent thermosensitivity that could be modulated by their native lipid environment. Interestingly, similar state-dependent salt bridges near the membrane interface have also been implicated in cold-activated TRP channels, including TRPM8 (Yin et al., 2022), raising the possibility that lipid-modulated electrostatic interactions may be a common thermodynamic strategy across various channel families. These findings highlight how detailed structural and functional studies of a bacterial model system can generate a conceptual framework for developing hypotheses about eukaryotic systems, for which similar experiments are currently not possible.

Using bacterial model systems to study the lipid dependence of ion channels is not limited to membrane-forming phospholipids but can be extended to common signaling lipids typically found in eukaryotes, such as PIP2. Functional experiments revealed that PIP2 efficiently inhibits SthK activity, similar to the effect observed in CNG channels (Womack et al., 2000; Thon et al., 2024). Structural studies of SthK under identical conditions to the functional experiments revealed a PIP2-binding site (Fig. 7 A), which was validated by mutational studies. In contrast to anionic membrane lipids, which electrostatically destabilize a salt bridge at the bundle crossing, PIP2 is positioned nearby, where it holds the S4 helix and the C-linker in place in the closed state, thereby sterically inhibiting channel opening by preventing conformational changes of the CNBD associated with channel opening (Fig. 7 A) (Thon et al., 2024). This binding site is conserved between SthK and CNG channels (Fig. 7, A and B) but is not present in HCN channels. As such, SthK is a natural chimeric protein that shares features with both eukaryotic CNG and HCN channels. Recently, the knowledge obtained from SthK was used to design targeted experiments to elucidate the mechanism of action of PIP2 on CNG channels (Park and Nimigean, 2025, Preprint). Remarkably, PIP2 also inhibited purified, liposome-reconstituted eukaryotic CNGA1 channels of the rod photoreceptors by binding at an equivalent location to SthK (Fig. 7), once again demonstrating the equivalence between model and eukaryotic homolog.

An intriguing aspect of this finding is that PIP2 is not a major component of bacterial membranes, raising the question of whether the binding site in SthK evolved to interact with a native lipid. One possibility is that phosphatidylinositol (PI), present in certain bacteria, including Spirochetes (Belisle et al., 1994), may occupy this site under physiological conditions. Although PI lacks the multiply phosphorylated headgroup of PIP2, it could still provide a structural scaffold for interaction with the channel, albeit with reduced electrostatic potential. In this view, PIP2 may not simply mimic a native ligand but instead stabilize a distinct modulatory state through its higher charge density and specific headgroup interactions. More broadly, this observation suggests that lipid interaction motifs present in prokaryotic channels may represent conserved structural frameworks that evolved to recognize specialized signaling lipids in eukaryotic systems.

Initial functional characterization of purified SthK revealed that its channel activity increases during prolonged incubation with cAMP, reminiscent of the slow cAMP-dependent current increase observed in HCN2 (Kusch et al., 2010; Schmidpeter et al., 2018; Schmidpeter et al., 2020a). Structurally, cAMP-dependent activation involves an upward swing of the C-helix to interact with the bound ligand, which is only possible if the siphon region, a short helix-turn-helix motif, also moves upward (Fig. 8 A) (Zagotta et al., 2003; Kesters et al., 2015; Lee and MacKinnon, 2017).

Interestingly, a proline residue is located at the end of the first siphon helix (Pro300 in SthK), a position conserved in all HCN channels but not in CNG channels (Fig. 8, A–C). A detailed study of cAMP-mediated activation of SthK showed that the cis/trans conformation at Pro300 in the siphon determines both the activation kinetics and the EC₅₀ for cAMP in SthK (Schmidpeter et al., 2020a; Newton et al., 2025). With trans-Pro300, activation is fast and the EC₅₀ for cAMP is low, whereas with cis-Pro300, activation is slow and the EC₅₀ is higher. Further analysis of WT SthK and SthK P300A (an all-trans mimic) indicated that in apo SthK, cis-Pro300 is favored. In contrast, in the open state, the cis/trans equilibrium at Pro300 is shifted almost completely toward the trans-Pro300 species (Fig. 8 D). Prolyl isomerases, enzymes that catalyze the cis/trans interconversion of prolyl bonds, are essential for enabling this regulation on a physiological timescale (Schmidpeter et al., 2020a).

The proline in the siphon is conserved between SthK and all HCN channels (Fig. 8, A–C). However, whether HCN channels are modulated by native-state prolyl isomerization remains elusive, largely due to a lack of functional assays for HCN channels that could resolve this mechanism. Nevertheless, this offers an additional example of how bacterial homologs, which can be readily purified and characterized using various functional in vitro assays, can serve as a basis for developing hypotheses about the regulation of their eukaryotic counterparts.

Despite their many advantages, prokaryotic model channels do not fully recapitulate the regulatory complexity of their eukaryotic counterparts. Eukaryotic ion channels are often subject to multiple layers of regulation, including interactions with accessory subunits, posttranslational modifications, and coupling to diverse signaling pathways. These features can profoundly influence channel behavior in ways that simplified bacterial systems do not capture. In addition, heteromeric assembly, subcellular localization, and dynamic lipid composition in native membranes introduce further layers of regulation that remain challenging to reproduce in vitro. Taken together, these considerations highlight that bacterial model channels are best suited to dissect specific, isolated regulatory mechanisms but may not capture the full extent of polymodal regulation found in many eukaryotic channels (MacKinnon and Doyle, 1997). Rather than serving as substitutes, bacterial homologs provide experimentally tractable systems that isolate core biophysical principles, enabling hypothesis generation and informing experimental design and mechanistic interpretation where direct approaches remain inaccessible.

In the early days of ion channel structural biology, bacterial channels provided seminal high-resolution structures that revealed the fundamental principles of K⁺ channels, which are conserved across all kingdoms of life. With the advent of high-resolution cryo-EM, solving the structures of complex eukaryotic channels has become more feasible. However, for many eukaryotic channels, detailed correlations between structural states and functional outcomes under identical experimental conditions remain challenging to establish. In this context, homologous bacterial ion channels can provide a valuable path toward detailed structure–function correlations.

In addition, the availability of pure, homogeneous preparations of simplified biophysical model systems, suitable for a range of assays from functional assays to high-resolution structural techniques, also enables more specialized discoveries for specific K⁺ channel subfamilies. In light of that, the role of bacterial ion channels has evolved from a necessary simplification for in vitro ion channel biophysics. They have become hypothesis-generating tools that can inform experimental design, presenting a compelling avenue for developing strategies to address otherwise inaccessible information, particularly when exploring challenging eukaryotic targets.

No new data were generated or analyzed in support of this work.

Brad Rothberg served as editor.

This work was supported by grants from the National Institutes of Health (R35GM159863 to P.A.M. Schmidpeter, and GM124451 and GM088352 to C.M. Nimigean).

Author contributions: Philipp A.M. Schmidpeter: conceptualization, data curation, formal analysis, funding acquisition, investigation, supervision, validation, visualization, and writing—original draft, review, and editing. Crina M. Nimigean: conceptualization, funding acquisition, investigation, and writing—review and editing.