Investigating Potassium Channels in Budding Yeast: A Genetic Sandbox

Abstract

Like all species, the model eukaryote Saccharomyces cerevisiae, or Bakers’ yeast, concentrates potassium in the cytosol as an electrogenic osmolyte and enzyme cofactor. Yeast are capable of robust growth on a wide variety of potassium concentrations, ranging from 10 µM to 2.5 M, due to the presence of a high-affinity potassium uptake system and a battery of cation exchange transporters. Genetic perturbation of either of these systems retards yeast growth on low or high potassium, respectively. However, these potassium-sensitized yeast are a powerful genetic tool, which has been leveraged for diverse studies. Notably, the potassium-sensitive cells can be transformed with plasmids encoding potassium channels from bacteria, plants, or mammals, and subsequent changes in growth rate have been found to correlate with the activity of the introduced potassium channel. Discoveries arising from the use of this assay over the past three decades have increased our understanding of the structure–function relationships of various potassium channels, the mechanisms underlying the regulation of potassium channel function and trafficking, and the chemical basis of potassium channel modulation. In this article, we provide an overview of the major genetic tools used to study potassium channels in S. cerevisiae, a survey of seminal studies utilizing these tools, and a prospective for the future use of this elegant genetic approach.

Potassium Channel Classes, Activities, and Structures

POTASSIUM channels are ubiquitous in eukaryotes and prokaryotes, and are the most diverse type of cation channel due to early evolutionary divergence (Loukin et al. 2005). The most salient feature of potassium channels is their ability to discriminate K⁺ from smaller cations such as Na⁺ (MacKinnon 2003). All potassium channels have fourfold symmetry around their central pore (MacKinnon 1991). This is typically accomplished by tetramerization of identical or homologous subunits, although K_2P channels are dimers in which two pore-forming domains are fused in tandem. The paradigmatic feature of all potassium channels is the eight-amino acid potassium selectivity motif, TXXTXG(Y/F)G. The alignment of the peptide backbone carbonyls of the selectivity motif mimic the hydration shell of a potassium (K⁺) ion, thus favorably dehydrating K⁺ ions while excluding ions with smaller van der Waals radii, such as Na⁺ or Li⁺ (Doyle et al. 1998; Roux and MacKinnon 1999).

Potassium channels are broadly classified based on the number of transmembrane domains (TMD): six-TMD voltage-gated (K_v) channels, four-TMD two-pore channels (K_2P), and two-TMD inward-rectifying (K_ir) channels (Miller 2000; Enyedi and Czirják 2010).

In voltage-gated channels—also known as Shaker-type channels after the first such channel identified in Drosophila melanogaster (Tempel et al. 1987)—the selectivity filter is situated between TMD5 and TMD6. In addition, TMD4 is distinguished by highly conserved arginine or lysine residues at every third or fourth position along the transmembrane helix, which are essential for sensing transmembrane voltage and channel opening upon membrane depolarization (Figure 1). Calcium-gated potassium channels (K_Ca) are also six-TMD-containing channels and have diverged from K_v channels, though some (notably “big current” or “BK” channels) retain voltage sensitivity. As their name implies, they are activated upon binding intracellular Ca²⁺ through various mechanisms (Vergara et al. 1998; Ghatta et al. 2006).

Figure 1.

Topological and structural features of major potassium channel clades. (A) Cartoon representation of the transmembrane topology and key functional features of several subclasses of potassium channel monomers mentioned in this review. Note the universally conserved pore domain containing two transmembrane domains (TMDs), a pore helix, and a selectivity filter. (B) Representative structures of mammalian Shaker (K_v1.2), a six-TMD channel, and (C) bacterial KcsA, a two-TMD channel, shown looking down the pore from the extracellular side (left images) and parallel to the membrane (right images). Both structures are homotetramers with individual polypeptides rendered in different colors. The Shaker structure is derived from Protein Data Bank identifier (PDB ID): 2A79 (Long et al. 2005), and the KcsA structure is derived from PDB ID: 2WLM (Clarke et al. 2010). All images were rendered in PyMol (v. 1.8.6).

Multicellular eukaryotes also express tandem pore domain channels, or K_2P channels, which appear to be the product of a gene fusion event (Figure 1). K_2P channels are responsible for very small “leak” currents that are voltage-insensitive and nonrectifying, yet crucial for maintaining resting membrane potential. However, K_2P channels can be gated by a wide variety of stimuli, such as membrane stretching or deformation, heat, and pH (Enyedi and Czirják 2010). These properties give rise to numerous specialized roles for K_2P channels in cells and tissues that sense physical stimuli. For example, the most highly studied K_2P channels—the TASK and TREK families—are involved in pain reception and are a major target of volatile anesthetics (Bayliss and Barrett 2008).

Inward-rectifying channels have a simpler structure of two TMDs flanking the selectivity filter (Hibino et al. 2010) (Figure 1). They are presumably the basal clade of potassium channels with widely distributed homologs, even in prokaryotic genomes (Loukin et al. 2005). Their bias for inward over outward currents (in apparent violation of the Nernst equation) occurs via transient block by intercellular divalent cations, primarily Mg²⁺ and polyamines (Lopatin et al. 1994). These divalent anions interact at multiple sites in the extended cytoplasmic pore of K_ir channels (Nichols and Lopatin 1997). However, in mammalian physiological settings, K_ir channels rarely encounter sufficient voltage to allow potassium influx and are usually efflux channels. The “classical” mammalian K_ir channels (K_ir2.x) are constitutively active—i.e., they have a high open probability (P_o) and are regulated primarily by the amount of protein at the cell surface—while other classes have a low P_o until activated by G-proteins (e.g., K_ir3.x) or intracellular nucleotides (e.g., K_ATP or K_ir6.x) (Hibino et al. 2010).

Based on their activities, regulation, and distribution, potassium channels mediate a broad range of biological processes. For example, in humans they are involved in nerve conduction (Mathie and Veale 2009), cardiac contraction (Sanguinetti and Tristani-Firouzi 2006), hearing (Wangemann 2006), maintenance of blood pressure (Wang et al. 2010), immune responses (Cahalan et al. 2001), and numerous other processes beyond the scope of this review. Moreover, polymorphisms in human potassium channels have been linked to dozens of genetic disorders affecting the nervous system, heart, kidneys, endocrine glands, and immune cells [see Kim (2014) for an extensive review of channelopathies]. Unfortunately, investigating potassium channel activity, modulation, and biogenesis is often labor-intensive, especially in mammalian cells. The development of a rapid, genetically tractable model in which these studies can be conducted has significantly aided our understanding of potassium channel physiology.

Potassium Channels and Potassium Homeostasis in the Yeast Saccharomyces cerevisiae

The budding yeast Saccharomyces cerevisiae presents several properties that make it an attractive system in which to probe the function of potassium channels. First, genetic transformations are famously facile in S. cerevisiae as compared to other commonly studied organisms. Second, strain libraries have been developed in which every gene in the yeast genome has been deleted, can be conditionally deactivated, or is present at reduced copy. Third, these unicellular eukaryotes can be propagated on either solid or in liquid growth medium, so yeast numbers are eminently scalable and the organism is amenable to high-throughput manipulations. Fourth, laboratory strains can be transitioned between the haploid and diploid state, allowing for genetic recombination and analyses of recessive and dominant phenotypes. Fifth, the resting membrane potential in S. cerevisiae is believed to be high relative to mammalian cells (see below), thus driving inward cation currents. Finally, despite nearly a billion years of evolutionary separation from humans, the vast majority of subcellular processes that animate yeast are conserved. Therefore, insights from yeast research almost always translate into multicellular eukaryotes.

Plasma membrane alkali metal transport in S. cerevisiae is adapted to efficiently import potassium and exclude sodium from the cytosol [for a detailed review, see Ariño et al. (2010)] (Figure 2). In brief, the electromotive force for potassium influx is generated by the electrogenic plasma membrane proton ATPase Pma1. Transmembrane potential in yeasts is believed to be quite high due to the activity of this proton transporter, but is notoriously difficult to measure directly with electrophysiological methods due to the high rigidity of the yeast cell wall (Vacata et al. 1981; Volkov 2015). Pma1 is the most abundant plasma membrane protein in yeast and consumes ∼20% of cellular ATP.

Figure 2.

Schematic of plasma membrane proton and alkali metal channels and transporters in S. cerevisiae. Native yeast channels and transporters mentioned in this text are shown here with representative ion currents. This figure has been adapted and simplified from previously published work (Ariño et al. 2010). Select phenotypes of the potassium channel and transporter deletion strains mentioned in the text are shown to the right (Bañuelos et al. 1998; Bertl et al. 2003) Channels deleted in the trk1Δ trk2Δ and trk1Δ trk2Δ tok1Δ strains are shown in orange, and transporters deleted in the ena1-4Δ nhx1Δ strain are shown in blue.

The Pma1-dependent transmembrane potential provides the electromotive force for potassium uptake through two high-efficiency channels, Trk1 and Trk2 (Figure 2). Trk1 was identified as the primary avenue for potassium influx (Gaber et al. 1988; Ko and Gaber 1991) and its structure is quite distinct from mammalian potassium channels. Most prominently, the pore-forming unit is a single polypeptide consisting of four two-TMD units, which together form a fourfold symmetric pore. Under most growth conditions, Trk1 is responsible for nearly all potassium influx, but it is not essential. Yeast lacking TRK1 (trk1Δ) have slightly increased potassium requirements that nevertheless fall within the bounds of ordinary laboratory growth medium. These yeast survive due to the presence of Trk2, which is a Trk1 paralog with a lower affinity for potassium and far lower expression in wild-type yeast. In contrast, doubly mutated trk1Δ trk2Δ yeast have a drastically increased requirement for potassium supplementation in growth medium, usually ∼100 mM. That these yeast are able to survive at all speaks to the presence of a very low-affinity potassium uptake system, termed “nonspecific cation channel 1” (Nsc1) (Bihler et al. 1998). The gene encoding this putative channel is as yet unidentified, and in fact may not even exist because potassium may leak into trk1Δ trk2Δ cells through various nutrient transporters. Indeed, mutations in hexose or amino acid transporters are a common cause of phenotypic reversion in trk1Δ trk2Δ yeast (Ko et al. 1993; Wright et al. 1997; Liang et al. 1998). This can be prevented by propagating trk1Δ trk2Δ cells on high-potassium medium up until the time that key experiments are conducted (see below). Nevertheless, trk1Δ trk2Δ cells are hyperpolarized and are sensitive to low pH (Ko and Gaber 1991), to alkali metals other than potassium (Bertl et al. 2003), and to a variety of toxic cations such as hygromycin B (Madrid et al. 1998).

In addition to the Trk channels, S. cerevisiae express a voltage-gated outward-rectifying potassium channel, Tok1. Like K_v channels in animals, Tok1 opens upon membrane depolarization. Tok1 is neither essential nor does deletion of TOK1 make trk1Δ trk2Δ yeast more sensitive to growth on low-potassium medium; however, compared to trk1Δ trk2Δ yeast, the triple-mutant strain exhibits increased tolerance to high cesium, which competes for potassium (Bertl et al. 2003). In addition, whereas mammals lack Trk1 or Trk2 homologs, Tok1 is homologous to the vertebrate two-pore K_2P channels (Wolfe and Pearce 2006). Therefore, select studies have utilized the trk1Δ trk2Δ tok1Δ potassium-sensitive strain (see below).

While potassium influx in yeast involves gradient-dependent, passive transport through the Trk1 and Trk2 channels, potassium efflux under conditions of excess cytosolic potassium requires active transport. Yeast have two primary systems to expel excess alkali metals into the extracellular environment: the ENA (exitus natru) uniporters and the Nha1 Na⁺/H⁺ antiporter (Figure 2). Although they primarily transport sodium, these systems are also able to export excess potassium (Bañuelos et al. 1998). ENA uniporters are encoded as tandem repeats at a single locus on chromosome IV with high (∼97%) conservation. The number of ENA isoforms varies among laboratory strains, with W3031-A having four copies and S288C having five. This may partially explain the observation that these two common laboratory workhorses differ in their tolerance to sodium and other cations (Petrezselyova et al. 2010). Deletion of the ENA cluster renders yeast sensitive to both sodium and lithium. ENA expression is typically low but is strongly induced upon exposure to high sodium or lithium via the action of the osmoresponsive transcription factor Hog1 (Märquez and Serrano 1996). In contrast, Nha1 is constitutively expressed at low levels, and seems to play a housekeeping role in regulating intracellular potassium under low pH and in expelling the small quantities of Na⁺ that enter through other transporters, such as the Na⁺/PO₄³⁻ symporter Pho89 (Persson et al. 1998). Both Nha1 and Ena1 promote growth in high-K⁺ medium, and a strain lacking both the ENA cluster (ena1-4Δ) and NHA1 fail to propagate on high concentrations of potassium or any other alkali metal (Bañuelos et al. 1998; Kinclová et al. 2001).

Yeast are also capable of sequestering alkali metals in organelles, particularly the vacuole. Vnx1 is a vacuole-localized antiporter that can exchange vacuolar protons for cytosolic sodium or potassium (Cagnac et al. 2007). Vnx1 likely plays an important role in detoxifying the cytosol upon sodium stress as vnx1Δ yeast are sensitive to high sodium (Cagnac et al. 2007) and the genomes of halotolerant fungi such as Hortaea werneckii contain multiple paralogs of VNX1 (Plemenitaš et al. 2016). While concentrating potassium in the vacuole can contribute to cellular turgor and help to balance the negative charge of vacuolar polyphosphate (Klionsky et al. 1990), it is unclear whether the vacuole can function as a “strategic reserve” upon acute potassium starvation.

The Awesome Power of the Potassium Growth Phenotype: Discoveries with trk1Δ trk2Δ Yeast

In this next section, we survey the many uses of trk1Δ trk2Δ yeast and then discuss several studies that have significantly aided our understanding of potassium channel function, regulation, biogenesis, and pharmacology. For the sake of brevity, we limit discussion to seminal studies and the most recent findings. For further information, we point the reader to the studies outlined in Table 1, which provides a more comprehensive list of discoveries in which exogenous potassium channels expressed in trk1Δ trk2Δ yeast played a key role.

Table 1. Select studies utilizing expression of an exogenous channel or transporter in trk1Δ trk2Δ, trk1Δ trk2Δ tok1, or ena1-4Δ nha1Δ yeast.

Channel	Primary method	Full citation
K2P2.1	Drug screen	Bagriantsev et al. (2013)
Kir2.1	Drug screen	Zaks-Makhina et al. (2004)
Kir2.1	Drug screen	Zaks-Makhina et al. (2009)
Kir3.2 (GIRK2)	Drug screen	Kawada et al. (2016)
Akt1	Expression	Sentenac et al. (1992)
AKT2	Expression	Cao et al. (1995)
HIV-1 Vpu	Expression	Herrero et al. (2013)
Hkt1	Expression	Schachtman and Schroeder (1994)
Kat1	Expression	Anderson et al. (1992)
Kat1	Expression	Schachtman et al. (1992)
Kir2.1	Expression	Hasenbrink et al. (2005)
Kir2.1 (IRK1)	Expression	Tang et al. (1995)
Kir6.1/6.2	Expression	Graves and Tinker (2000)
Kv11.1 (hERG)	Expression	Schwarzer et al. (2008)
ORK1	Expression	Goldstein et al. (1996)
RhAG (not a K channel)	Expression	André et al. (2000)
SKD1,2 (Transporters)	Expression	Périer et al. (1994)
TcCat	Expression	Jimenez and Docampo (2012)
AKT1	Random mutagenesis	Ros et al. (1999)
Hak5	Random mutagenesis	Alemán et al. (2014)
HvHAK	Random mutagenesis	Mangano et al. (2008)
Kat1	Random mutagenesis	Nakamura and Gaber (2009)
Kat1	Random mutagenesis	Nakamura et al. (1997)
Kat1	Random mutagenesis	Lai et al. (2005)
KcsA	Random mutagenesis	Paynter et al. (2008)
Kir3.1	Random mutagenesis	Sadja et al. (2001)
Kir3.2	Random mutagenesis	Bichet et al. (2004)
Kir1.1 (ROMK)	Random mutagenesis	Paynter et al. (2010)
Kir2.1	Random mutagenesis	Cho et al. (2000)
Kir2.1	Random mutagenesis	Minor et al. (1999)
Kir3.2 (GIRK2)	Random mutagenesis	Yi et al. (2001)
Kcv	Random mutagenesis	Chatelain et al. (2009)
K2P2.1 (TREK-1)	Random mutagenesis	Bagriantsev et al. (2011)
Kir2.1	Random mutagenesis	Chatelain et al. (2005)
TRVP4	Random mutagenesis	Loukin et al. (2010)
Kcv	Synthetic biology	Cosentino et al. (2015)
Kir1.1 (ROMK)	Synthetic genetic array	Mackie et al. (2018)
Kir2.1	Synthetic genetic array	Kolb et al. (2014)
Kir2.1	Trafficking assays	Grishin et al. (2006)
Kir2.1	Trafficking assays	Bernstein et al. (2013)
K2P3.1 (TASK-1)
K2P9.1 (TASK-3)
Kir3.2 (GIRK2)	Trafficking screening	Bagriantsev et al. (2014)
Kir6.1	Trafficking assays	Zerangue et al. (1999)
Kcv, Kesv	Trafficking assays	Balss et al. (2008)
Kir1.1 (ROMK)	Trafficking assays	O’Donnell et al. (2017)
Kir2.1	Trafficking screening	Shikano et al. (2005)

The identification of uncharacterized potassium channels from cDNA libraries

The Gaber laboratory first noted the potential of exogenous potassium channels to substitute for the endogenous potassium uptake system in yeast (Ko and Gaber 1991; Nakamura and Gaber 1998). Numerous studies over the past two decades have exploited this phenotypic assay to identify novel classes of potassium channels from various organisms. By transforming cDNA libraries into trk1Δ trk2Δ yeast and identifying clones that enable growth on low-potassium media, researchers have identified new potassium channels rapidly and with minimal bias. This technique offers several advantages over traditional screens for new channels. In this case, cRNA libraries are injected into Xenopus laevis oocytes and then ion current is measured. This labor-intensive protocol led to the identification of many potassium channels [see for example Ho et al. (1993)]. In contrast, the yeast selection scheme is cheaper, there is no need for electrophysiology “rigs,” and the isolation of positive clones does not require sequential subdivision of clonal oocyte pools and the isolation and amplification of unique clones because transformed yeast only express single clones. This technique has provided a particular boon for identifying potassium channels as well as transporters from the plant kingdom.

Potassium is a key macronutrient for plants that is absorbed from the soil through the root system and concentrated in cells to maintain cellular turgor. Because plants lack contractile tissues, potassium fluxes underpin various tropisms that allow them to respond to environmental stimuli (Dreyer and Uozumi 2011). The first plant potassium channel, KAT1, was identified by coopting the trk1Δ trk2Δ growth phenotype on low potassium (Anderson et al. 1992; Schachtman et al. 1992). KAT1 is a Shaker-type voltage-gated channel with inward-rectifying properties, is highly expressed in guard cells, and its activation in response to hyperpolarization causes the stoma to swell and open. Dozens of subsequent studies have since identified previously unknown potassium channels and transporters in both model and crop plants, including AKT1, a strongly inward-rectifying potassium uptake channel in root hairs (Sentenac et al. 1992; Hirsch et al. 1998), as well as HKT, which was the first plant high-affinity potassium transporter identified (Schachtman and Schroeder 1994) (see Table 1). Potassium transporters are functionally and morphologically distinct from potassium channels and, as such, fall outside the scope of this review, but they have also been extensively studied in yeast [see Rodríguez-Navarro and Rubio (2006) for review].

One limitation of this system bears mention. Full-length K_v channels from animals are unable to rescue the growth of trk1Δ trk2Δ yeast on low potassium. This phenomenon arises due to the fact that mammalian K_v channels fail to open in the hyperpolarizing potential across the yeast membrane; in contrast, plant Kv channels open under these conditions, which is why KAT1 and other related channels could be isolated in yeast (Table 1). However, the K_v channel rat ether-à-go-go-related gene-1 (rEAG1) was able to restore growth in trk1Δ trk2Δ cells if the N-terminal 190 amino acids (i.e., the entire cytosolic N-terminal domain) were truncated (Schwarzer et al. 2008). This result, while potentially opening many doors for studying mammalian K_v channels in yeast, is puzzling as the N-terminal PAS domain of hERG, the human homolog of rEAG1, is necessary for channel activity in native systems (Gianulis and Trudeau 2011). Fortunately, many mammalian K_ir channels function in trk1Δ trk2Δ and ena1-4Δ nha1Δ yeast [e.g., Graves and Tinker (2000), Yi et al. (2001), and Paynter et al. (2010), and also see Table 1]. The first of these—and to date the most extensively studied—is the cardiac inward-rectifier K_ir2.1, which was originally called IRK1 (Tang et al. 1995) (see below).

New potassium channels can currently be predicted by in silico recognition of core features based on DNA sequence analysis. Nonetheless, trk1Δ trk2Δ yeast remain a valuable tool to validate the function of putative potassium channels detected in the ever-expanding collection of sequenced genomes. This approach is particularly useful for exotic potassium channels that lack the “canonical” potassium selectivity motif. Many of these noncanonical channels have been found in the genomes of viruses or intracellular parasites, such as Trypanosoma cruzi (Jimenez and Docampo 2012).

Viroporins are small transmembrane proteins encoded in viral genomes that insert into the host cell plasma membrane, increase cell permeability, and play important roles in virion maturation and budding (Nieva et al. 2012). Because they are exposed to the extracellular environment, viroporins are also attractive therapeutic targets. Expression of viroporins from various viral taxa in trk1Δ trk2Δ yeast established that these proteins function as potassium channels. Notable viroporins studied in this system include Paramecium bursaria Chorella virus-1 Kcv (Chatelain et al. 2009), human immunodeficiency virus-1 protein U (Vpu) (Herrero et al. 2013), and porcine epidemic diarrhea virus ORF3 (Wang et al. 2012). Vpu lacks a canonical potassium selectivity motif, has a pentameric (rather than a tetrameric) pore structure, and only weakly selects potassium over other small monovalent cations, such as sodium and ammonium (Kukol and Arkin 1999). In contrast, a potassium selectivity filter was predicted in Kcv based on its presence in a viral genome (Plugge et al. 2000), even though Kcv is the smallest canonical potassium channel identified (each subunit is only 94 amino acids) and completely lacks a cytosolic C-terminal domain (Figure 1A) (Gazzarrini et al. 2003). While Kcv forms a functional ion channel at the host plasma membrane, Kesv, a similar viroporin from a related virus, is sorted to the mitochondria. Expression in trk1Δ trk2Δ yeast showed that Kesv mitochondrial targeting is dependent on the length of TMD2 and that a slight extension of TMD2 retargets the channel to the secretory pathway (Balss et al. 2008). Therefore, yeast expression studies have elucidated some unexpected features of viral potassium channels.

Structure–function relationships via mutagenesis and genetic selection

Random mutagenesis has long been used as a rapid and nonbiased method to identify critical residues, motifs, and domains in ion channels. The approach may target specific regions, in which degenerate primer sets scramble a small segment of the coding sequence, or the entire coding sequence or a sizable portion thereof through error-prone PCR. Gene shuffling can also be used to efficiently mutagenize large genes and to perform in vitro backcrosses to reduce the number of mutations per clone, or to ensure the presence of a particular mutation in a screen for second-site suppressors (Stemmer 1994; Bichet et al. 2004). These studies are easily undertaken in yeast as transformed yeast cells only uptake single clones from which mutations can be quickly identified. In addition, homologous recombination can be used to join multiple linear DNA fragments, one of which can contain a mutagenized insert (created, for example, by PCR) and one that contains the expression plasmid via direct transformation into recipient yeast strains (Weir and Keeney 2014).

Targeted mutagenesis studies of KAT1 focused on the selectivity filter. The conserved “YG” site in the canonical selectivity filter TXXTXGYG was mutated to 400 possible combinations using a degenerate primer set (Nakamura et al. 1997). While some conservative mutations of the tyrosine produced nominally functional channels with impaired selectivity (i.e., growth on low potassium in the absence of sodium), only the most conservative YG→FG mutation resulted in a fully selective channel that allowed growth of trk1Δ trk2Δ yeast on low-potassium media in the presence of high sodium. When the residues flanking the selectivity motif were instead mutated, multiple mutations increased Na⁺ exclusion, as evidenced as well by increased growth on low potassium as well as high sodium. Curiously, the mutations in the flanking region, which led to increased Na⁺ exclusion, also allowed the transport of large toxic cations such as Cs⁺ and triethylammonium that normally block the channel. These data suggest that the entire pore region has evolved under dual selective pressure to exclude cations both larger and smaller than K⁺ (Nakamura and Gaber 2009).

While residues conferring selectivity are confined to a small but highly conserved region in potassium channels, residues that regulate gating are often more broadly distributed. Jan and colleagues performed mutagenesis to determine the importance of the voltage-sensing TMD4 helix in relation to the pore forming domain in TMD5 in KAT1. Mutations in TMD5 that conferred a mild loss-of-function phenotype were isolated and then random mutagenesis of TMD4 uncovered second-site suppressors, thus providing evidence of the close proximity of TMD4 and TMD5, particularly when the channel is in the open state (Lai et al. 2005). Similar studies from the same laboratory focused on identifying the gating mechanisms of mammalian K_ir channels. Specifically, TMD1 and TMD2 in K_ir2.1 were randomly mutagenized and, based on the relative effects of the mutations on channel function in trk1Δ trk2Δ yeast, the authors predicted the lipid-facing, pore-facing, and helix-packing surfaces of the TMDs (Minor et al. 1999). These predictions would later be validated by an analysis of the crystallographic structure of a bacterial inward-rectifying channel, KirBac1.1, in the closed state (Kuo et al. 2003).

Of the mammalian K_ir channels, K_ir2.1 is the most amenable to studies in trk1Δ trk2Δ yeast as it is highly expressed, traffics efficiently to the plasma membrane, and its P_o is high (Minor et al. 1999). A random mutagenesis screen in trk1Δ trk2Δ yeast for K_ir2.1 channels resistant to barium block indicated that the dipole moment of the pore helix does not play a major role in stabilizing cations in the pore, as barium-resistant mutants with positively-charged substitutions at the C-terminus of the pore helix did not exhibit reduced single-channel potassium current (Chatelain et al. 2005).

Unlike K_ir2.1 and other classical K_ir channels, the G-protein-activated inward-rectifying potassium channel-2 (GIRK2, also known as K_ir3.2) has a much lower P_o (∼0.01) unless activated by G_βγ binding. A random mutagenesis screen of the entire GIRK2 coding sequence identified five residues in the pore domain that are necessary for gating, and substitution at any of these positions increased the P_o independently of G-protein stimulation. The authors also identified a specific valine in TMD2 that plays a critical role during channel opening (Yi et al. 2001). A similar analysis performed contemporaneously by another group also found pore domain mutations necessary for GIRK1 (K_ir3.1) channel gating in the absence of G-proteins (Sadja et al. 2001). A follow up to the initial mutagenesis of K_ir3.2 utilized gene shuffling to find second-site suppressors of the constitutively active but selectivity-deficient S177W mutation. This study identified several mutations in the TMDs that were able to functionally restore potassium selectivity (Bichet et al. 2004).

K_2P channels can be gated by a much wider variety of physical and chemical stimuli than either K_ir or K_v channels. For example, the highly-studied human TREK-1 (K_2P2.1) channel is gated by both intra- and extracellular low pH, heat, mechanical membrane deformation, and anesthetics (Enyedi and Czirják 2010). Random mutagenesis of TREK-1 and expression in trk1Δ trk2Δ yeast revealed that TREK-1’s response to these disparate stimuli relies on a common gating mechanism. Specifically, a well-conserved extracellular region of TMD4 containing multiple aromatic residues was shown to be necessary for channel gating in TREK-1 and other K_2P channels (Bagriantsev et al. 2011).

Studies on potassium channel trafficking in the yeast model

The whole-cell potassium conductance, which determines the potassium current for a given driving force, is the product of the single-channel conductance, the open probability, and the total number of channels at the cell surface. Therefore, cells must tightly control the plasma membrane residence of ion channels by regulating synthesis, intramembrane trafficking, and degradation. Like all classes of transmembrane proteins, potassium channels interface with the intramembrane trafficking machinery through specific amino acid motifs (Griffith 2001). Most of these trafficking signals are short linear motifs, similar to canonical signals shared by a wide variety of membrane proteins, but they can also be found at the interface between two subunits (Ma et al. 2011; Li et al. 2016). Endoplasmic reticulum (ER)-retention motifs are short linear motifs found on the cytosolic face of transmembrane proteins and promote retrograde transport from the Golgi apparatus to the ER, thereby preventing cargo proteins from reaching the cell surface. One such sequence is the di-arginine “RXR” motif that interacts directly with the coatamer (COPI) (Zuzarte et al. 2009; Okamoto and Shikano 2011) and becomes occluded as the channel folds and assembles (Zerangue et al. 1999). RXR was first characterized in K_ir6.2 and is present in many mammalian K_ir channels. Channels that lack this motif, such as K_ir2.1, are robustly trafficked to the plasma membrane in both yeast and mammalian cells (Grishin et al. 2006). To identify novel ER exit/forward trafficking signals that may override RXR, Shikano and colleagues fused a random peptide library to K_ir2.1-RXR and screened for increased channel activity in trk1Δ trk2Δ yeast. As a result of this analysis, the authors identified SWTY as a 14-3-3 protein interactor (Shikano et al. 2005). 14-3-3 proteins bind to phospho-serine/threonine motifs to regulate myriad processes (Muslin et al. 1996; Kumar 2017). In both yeast and multicellular eukaryotes, the -2 threonine of the SWTY motif must be phosphorylated to bind 14-3-3 proteins, and this phenomenon was demonstrated using both model substrates as well as naturally occurring proteins with SWTY-like sequences (Shikano et al. 2005). In a similar study, novel modulators of GIRK2 were identified by fusing either a peptide library that contained random 15 amino acid motifs flanked by a well-folded domain from Staphylococcus aureus protein Z, or a randomly mutagenized version of protein Z itself to either the N- or C-terminus of GIRK2. This multipronged approach made for a more comprehensive screen by preventing bias against poorly folding peptides that are nonetheless strong positive regulators of GIRK2. As a result of this study, the authors identified a peptide that augments GIRK2 current by increasing plasma membrane residence (Bagriantsev et al. 2014).

In contrast to the identification of embedded sequences that regulate potassium channel trafficking, the yeast screen has also been used to find extragenic regulators of trafficking. This was accomplished through the use of synthetic genetic array (SGA) strategies (Tong and Boone 2006) in which a genome-wide analysis can be performed to identify genes that, when deleted, alter the growth of potassium channel-expressing trk1Δ trk2Δ yeast on low-potassium media (Figure 3). Based on the strength of the phenotype, the candidates can also be ranked.

Figure 3.

Visual representation of Synthetic Gene Array (SGA) technology to screen for potassium channel regulators. (1) The “magic marker” yeast strain, which contains the can1Δ marker for dominant selection against diploid cells and HIS3 under the control of the STE2 promoter to select for MATa haploids, is additionally deleted for trk1Δ and trk2Δ. This strain is then transformed with a potassium channel (“K-Chnl”) and mated to the yeast deletion library (“xxxΔ”). (2) Yeast colonies are transferred to selective media to propagate diploids, then onto low-nitrogen media to induce sporulation. (3) Sporulated yeast are grown on a series of selective media to isolate MATa haploid progeny harboring trk1Δ trk2Δ and the library hypomorph. All growth media in steps 1–3 should be supplemented with 100 mM K⁺ to prevent intergenic suppression of potassium sensitivity. (4). Triple-mutant (“trk1Δ trk2Δ xxxΔ”) yeast are transferred to high- and low-potassium media and growth is assessed. Adapted from Kolb et al. (2014) and the materials and methods section in Mackie et al. (2018).

The first instance in which an SGA was performed in yeast expressing a potassium channel used a targeted library of deletion strains lacking components associated with the early secretory pathway. The library was used to query yeast expressing a constitutively active GIRK variant (K_ir3.2_S177W) that increases Na⁺ influx (Yi et al. 2001). As a result of this phenotype, yeast growth was slowed on high-sodium media. Therefore, isolated strains with increased growth on high sodium were hypothesized to lack a factor that promotes K_ir3.2_S117W biosynthesis and forward trafficking. The genes identified in this analysis were primarily involved in sphingolipid biosynthesis, which may facilitate channel opening at the plasma membrane (Epshtein et al. 2009), and COPII vesicle cargo packing of GPI-anchored proteins, which may indirectly impact ER protein export. These data indicate a preeminent role of these pathways in K_ir channel biogenesis and function (Haass et al. 2007).

Subsequent studies from our group have applied an SGA to identify and rank regulators of potassium channels with the entire nonessential gene library, which contains > 5000 gene knockout strains. For this analysis, TRK1 and TRK2 were deleted in the “magic-marker” SGA query strain (Tong and Boone 2006), which resulted in slow growth on low-potassium media. The strain was then transformed with a plasmid encoding K_ir2.1 and mated to each strain in the deletion collection. Yeast displaying increased growth on low-potassium media were hypothesized to lack negative regulators of K_ir2.1. The result of this screen led to the isolation of yeast lacking genes encoding factors involved in vacuolar protein sorting (VPS) and several subunits of the endosomal complexes required for trafficking (ESCRT) (Kolb et al. 2014). These data suggested that K_ir2.1, while primarily being targeted for destruction by the ER-associated degradation (ERAD) pathway (see below), is also routed to the vacuole for degradation; therefore, slowed vacuole-dependent degradation liberated additional K_ir2.1 channels to the plasma membrane, which facilitated growth on low-potassium media. Consistent with this hypothesis, inhibition of the ESCRT pathway via knockdown strategies in human cells increased the steady-state pool of K_ir2.1.

More recently, we published on a related screen for regulators of the renal outer medullary potassium channel ROMK (also known as K_ir1.1). ROMK was the first K_ir gene identified in mammals (Ho et al. 1993) and is primarily expressed in the distal nephron, where it serves as a master regulator of extracellular potassium in response to changes in dietary intake. ROMK also plays an important role in sodium balance and the urinary concentrating mechanism (Welling and Ho 2009). ROMK is inhibited when the intracellular pH drops below 6.0 (Paynter et al. 2010), which necessitated the mutation of a pH-titratable lysine to a methionine (K80M) for functional studies in trk1Δ trk2Δ yeast. Unlike K_ir2.1, ROMK primarily resides in the ER at steady-state until Serine 44 becomes phosphorylated (Yoo et al. 2005), so a phosphomimic was next introduced at position 44, converting a serine to an aspartic acid (O’Donnell et al. 2017). An SGA screen with a ROMK-expressing query strain revealed that many of the same VPS genes responsible for removing K_ir2.1 from the plasma membrane similarly regulated ROMK, but an added contribution of the class C core vacuole/endosome tethering (CORVET) complex was evident (Mackie et al. 2018). CORVET is a heterohexameric complex that acts upstream of ESCRT by tethering postendocytic vesicles to form early endosomes (Balderhaar et al. 2013). As neither K_ir2.1 nor ROMK is native to yeast, it is likely that endosomal sorting represents a conserved mechanism to regulate K_ir channels. As for K_ir2.1, surface expression of ROMK increased when ESCRT subunits were silenced in cultured mammalian cells (Kolb et al. 2014; Mackie et al. 2018).

As mentioned in Potassium Channel Classes, Activities, and Structures, mutations in potassium channels give rise to a range of human diseases, and ROMK is no exception. Over 50 loss-of-function mutations in ROMK have been linked to the autosomal recessive salt-wasting disease Bartter Syndrome Type II (Welling and Ho 2009). We hypothesized that many of these mutants compromise ROMK folding or tetramerization, which would trap the channels in the ER and target them for ERAD. Yeast and multicellular eukaryotes utilize ERAD to recognize and degrade misfolded proteins in both the ER lumen and the ER membrane (Vembar and Brodsky 2008). As the ER lumen lacks a major degradative protease, the AAA+ ATPase Cdc48/p97 retrotranslocates substrates to the cytosol for degradation in the proteasome (Ye et al. 2001). Mutations in proteins that enter the secretory pathway, including several classes of ion channels, have been linked to many genetic diseases and lead to protein misfolding and ERAD targeting (Guerriero and Brodsky 2012). Indeed, we found that select Bartter Syndrome mutations in the C-terminus of ROMK cause a loss-of-function phenotype in trk1Δ trk2Δ yeast by increasing the ERAD of these mutants, thus establishing Bartter Syndrome Type II as yet another ERAD-linked channelopathy (O’Donnell et al. 2017).

Pharmacological studies of potassium channels in yeast

As a large and functionally multifaceted class of proteins, potassium channels represent a wellspring of potential targets for new additions to the pharmacopoeia (Kaczorowski et al. 2008; Wulff et al. 2009). Drugs that regulate potassium channels have proved efficacious in clinical settings. For example, rosiglitazone, which stimulates insulin release from pancreatic β-cells through inhibition of K_ATP (K_ir6.1), has been used to treat diabetes and hypertension (Yu et al. 2012), while retigabine, a potentiator of K_v7.2/K_v7.3 channels, has shown promise as an antiepileptic (Ihara et al. 2016). Conversely, the human ether-à-go-go-related gene (hERG or K_v11.1) is a common nemesis of drug trials as it is inhibited by many structurally and functionally unrelated compounds (Priest et al. 2008). Therefore, there is an unmet need for drugs that target potassium channels that have been heretofore refractory to medicinal chemistry or that exhibit fewer side effects. To these ends, potassium channel pharmacology lends itself naturally to the trk1Δ trk2Δ system as molecules can be added directly to growth media to interact with the plasma membrane pool of potassium channels. As many of these compounds interact with the extracellular face of the channel pore, their efficacy is unlikely to be affected by S. cerevisiae’s well-known arsenal of multidrug resistance ABC transporters.

In pioneering work, a library of > 10,000 synthetic compounds was screened for their ability to inhibit K_ir2.1-mediated rescue of trk1Δ trk2Δ yeast, as determined by the optical density of cultures in low-potassium medium. From a list of preliminary hits, compounds were excluded that inhibited other potassium channels, such as GIRK2 or KAT1. Ultimately, a synthetic compound was found that blocked the ion-conducting pore in K_ir2.1 and protected against potassium efflux-mediated caspase activation and apoptosis in primary cultured neurons (Zaks-Makhina et al. 2004). A follow-up study using a natural products library identified gambogic acid as a potent and long-lasting inhibitor of K_ir2.1 that alters the interaction of the channel with the membrane microenvironment (Zaks-Makhina et al. 2009). These screening techniques were then modified to find regulators of the mechano- and thermosensitive K_2P channel TREK-1 (K_2P2.1). K_2P channels are of broad interest in medicine due to their roles in pain, protection from ischemia, migraine, and depression. By screening a library of > 100,000 compounds in trk1Δ trk2Δ yeast that expressed TREK-1 and were grown in a variety of K⁺ concentrations, one compound was identified that activates TREK-1 and its close relatives, but not more distally related K_2P channels or other classes of potassium channels (Bagriantsev et al. 2013).

In another successful application of this technology, the KCNJ6 gene was used. KCNJ6, which encodes GIRK2, is located on chromosome 21 in humans, and its duplication in trisomy-21 (Down syndrome) has been linked to cognitive impairment (Rachidi and Lopes 2007). In a screen of > 2000 compounds in GIRK-expressing trk1Δ trk2Δ tok1Δ yeast, proflavine and several derivatives were identified as voltage-dependent blockers of GIRK2 (Kawada et al. 2016). It should be noted that the use of trk1Δ trk2Δ tok1Δ rather than trk1Δ trk2Δ allowed the authors to employ Cs⁺ as a positive control for GIRK2 channel blockage. In each of the other drug screens, a more pleiotropic positive control for cell death was used, such as SDS. In contrast, Cs⁺ specifically targets the exogenously expressed potassium channels in trk1Δ trk2Δ tok1Δ yeast.

Synthetic Biology and the Future of Potassium Channel Research in Yeast

Potassium channels and other ion channels are critical to maintain homeostasis and allow for information transduction and stimuli responses on much shorter timescales than any other biological process (Isacoff et al. 2013). Due to their ion-conducting properties and selectivity, potassium channels have attracted significant attention among synthetic biologists, who have endeavored to engineer novel gating mechanisms in potassium channels. Screens for neofunctionalized channels in yeast allow synthetic biologists to benefit from many of the same high-throughput molecular evolution techniques described thus far. For example, voltage sensitivity can be engineered into the Kcv channel upon fusion with a K_v-like voltage-sensing domain from a phosphatase (Arrigoni et al. 2013). More recently, a blue light-activatable potassium channel was created by fusing the light-sensitive LOV2 module from an Avena sativa phototropin to the N-terminus of Kcv. Error-prone PCR mutagenesis was then used to identify clones that rescue growth of trk1Δ trk2Δ yeast on low potassium when activated by blue light. The researchers then expressed this new potassium channel, termed blue light-induced potassium channel-1 (BLINK1), in Zebrafish embryos and found that their escape response reflex in response to touch was fully and reversibly inhibited upon exposure to blue light (Cosentino et al. 2015). This study represents the first instance of an optogenetically controlled potassium channel and will enable the creation of transgenic animals expressing photo-activatable potassium channels in specific neurons.

In the future, we envision continued use of the yeast screen for therapeutics, as many channel modulators lack desired specificity. Drugs are also needed for other potassium channels, particularly those whose function increases the severity of various cancers (Huang and Jan 2014). Moreover, we envision screens for modulators of mutated potassium channels for which treatments are currently lacking. For example, the expression of Bartter Syndrome mutant forms of ROMK, which are trapped in the early secretory pathway and degraded by ERAD (see above), might respond to small-molecule protein folding correctors. In fact, corrector molecules have proved to be efficacious treatments for other disease-causing mutations in ion channels targeted for ERAD, such as CFTR (Van Goor et al. 2011). A small molecule screen might uncover pharmacological chaperones that promote the folding of mutant ROMK and allow it to escape ERAD and other forms of protein quality control.

In sum, recent studies from multiple laboratories, spanning a wide breadth of disciplines, have demonstrated the continuing utility of a technique originally developed a quarter century ago. From the earliest identification of novel potassium channels before the genomic era to frontiers in molecular medicine and biotechnology, yeast screens have contributed significantly to our understanding of potassium channel biology. To date, a scan of the potassium channel literature demonstrates that most investigations have focused on small group of channels—KAT1, K₂P_2.1, Kcv, K_ir2.1, and K_ir3.x—which are either isolated from well-studied model organisms or are intimately linked to human biology. However, this list barely scratches the surface of this diverse and ubiquitous class of biomolecules. As potassium channel science continues to advance, yeast-based assays will undoubtedly continue to play an important part.