Yeast gain-of-function mutations reveal structure–function relationships conserved among different subfamilies of transient receptor potential channels

Abstract

Transient receptor potential (TRP) channels found in animals, protists, and fungi are primary chemo-, thermo-, or mechanosensors. Current research emphasizes the characteristics of individual channels in each animal TRP subfamily but not the mechanisms common across subfamilies. A forward genetic screen of the TrpY1, the yeast TRP channel, recovered gain-of-function (GOF) mutations with phenotype in vivo and in vitro. Single-channel patch-clamp analyses of these GOF-mutant channels show prominent aberrations in open probability and channel kinetics. These mutations revealed functionally important aromatic amino acid residues in four locations: at the intracellular end of the fifth transmembrane helix (TM5), at both ends of TM6, and at the immediate extension of TM6. These aromatics have counterparts in most TRP subfamilies. The one in TM5 (F380L) aligns precisely with an exceptional Drosophila mutant allele (F550I) that causes constitutive activity in the canonical TRP channel, resulting in rapid and severe retinal degeneration beyond mere loss of phototaxis. Thus, this phenylalanine maintains the balance of various functional states (conformations) of a channel for insect phototransduction as well as one for fungal mechanotransduction. This residue is among a small cluster of phenylalanines found in all known subfamilies of TRP channels. This unique case illustrates that GOF mutations can reveal structure–function principles that can be generalized across different TRP subfamilies. It appears that the conserved aromatics in the four locations have conserved functions in most TRP channels. The possible mechanistic roles of these aromatics and the further use of yeast genetics to dissect TRP channels are discussed.

Transient receptor potential (TRP) channels are named after the near-blind mutant flies that show a TRP in their electroretinogram, first examined in the 1960s and 1970s (1–3). This work, together with the additional discoveries of related channels underlying sensitivity to heat (4) and mechanical forces (5) in 1997, kindled great interest in this class of channels. Today, >100 TRP channels have been reported in various animal models and in man, individually associated with the sensation of heat, cold, noxious chemicals, pain, osmotic force, touch, vibration, proprioception (6, 7), and axon guidance (8). Each TRP channel is considered a tetramer of subunits with six transmembrane (TM) α-helices. Animal TRPs are sorted into seven subfamilies (TRP-C, -A, -M, -ML, -N, -P, and -V) by distinguishing features in the predicted cytoplasmic domains preceding or trailing the TMs (6). Only the amino acid sequence from TM5 to just beyond TM6 is conserved among the subfamilies. TRPs with this sequence are found in the genomes of Paramecium, Tetrahymena, Dictyostelium, Trypanosoma, Leishmania, and other protists, as well as the two major fungal divisions, Basidiomycete and Ascomycete (9, 10).

TRPY1, the TRP channel of the budding yeast Saccharomyces cerevisiae, expresses in the vacuolar membrane (11), where it opens to release vacuolar Ca²⁺ into the cytoplasm when yeast cells are confronted by an osmotic upshock (12). Corresponding to this role in vivo, TRPY1 was found to be mechanosensitive under patch clamp (13). Like its counterparts in animals, TRPY1 is polymodal and also can be activated by cytoplasmic Ca²⁺, likely amplifying its own activation through a Ca²⁺-induced Ca²⁺-release feedback (13).

To understand TRP channels in general, we have been developing the study of TRPY1 (Yvc1p) (10, 11, 13) to complement animal studies. Single-molecule activities of TRPY1 can be examined in situ by direct patch clamping without the complexity of anatomy or heterologous expression encountered in animal research. The 320-ps unitary conductance of TRPY1 also offers a better signal-to-noise ratio in electric recording than the 20- to 60-ps animal TRPs. Perhaps the greatest advantage of the yeast system is its powerful genetics. Single-gene mutations in genomes have led to the identification of such key channels as Shaker (14), mec-4 (15, 16), CFTR (17), and various TRPs (5, 9). In theory, forward genetics can be taken one step further. Single-residue mutations upon random mutagenesis in a single-channel gene should allow the identification of key residues in channel functions. In practice, this type of comprehensive forward genetics focusing on a single gene is only practical with microbes or cultured cells, but not yet with plants and animals. Chief among the obstacles is the inability to transform large animal and plant populations en masse. We recently selected 10 TRPY1's gain-of-function (GOF) mutants after randomly mutagenizing the yeast's TRPY1 gene (not the entire genome). One such mutation shows that a tyrosine at the cytoplasmic end of TM6 is apparently needed to anchor the channel gate (18). Here we describe other GOF mutations, mainly those found in or near predicted TM domains, which all turned out to involve aromatic residues. One of these mutations coincides precisely with a unique GOF mutation in the canonical Drosophila TRP (19, 20).

Results

The GOF Screen.

The release of Ca²⁺ from the vacuole through TRPY1 upon osmotic upshocks (12) is gauged by the luminescence of transgenic aequorin (21). We randomly mutagenized YVC1 on a plasmid and screened for mutants among the transformed yvc1Δ yeast cells, which, upon the addition of an osmoticum, gave strong responses (the GOF phenotype). The plasmid from each of the candidates passing rescreens was retrieved, sequenced, restricted, and inserted into a new plasmid for yeast gene replacement. Yeast integrants with mutated YVC1 replacement were individually sequence-verified and examined by luminometry and patch clamp (19). All results described here are from these integrants. Ten such GOF mutants were selected for further analysis. Six strong mutations (F247L, S297W, F380L, Y458H, Y473H, and Y473C) are predicted to be in TM helices or TM6's immediate extension. Y458H, which affects the greatest gate destabilization, is described in Zhou et al. (18). The remaining five are analyzed here. Y442C, a weaker GOF mutant, also is analyzed here because of its location and conservation. For brevity's sake, we do not show data on the other three GOF mutations, which are located in the cytoplasmic domains, do not involve aromatics, and have no clear counterparts in nonfungal TRPs (18).

The GOF Mutations.

All of the 14 TM-assignment programs used predict Yvc1p to have six TMs, but with some disagreements in the demarcations (see Material and Methods). Based on the consensus, the F247L mutation is located in the middle of TM1, S297W in the cytoplasmic end of TM3, F380L in the cytoplasmic end of TM5, Y442C in the vacuolar end of TM6, Y458H in the cytoplasmic end of TM6, and Y473-H and −C in an amphipathic helical domain trailing TM6 (Fig. 1Top). Among these 10 GOF mutations, 6 replace aromatic residues and 1 (S297W) adds an aromatic. Our mutagenesis with the XL-Red mutator bacteria shows a bias of 34% each of A-to-G or T-to-C transitions among the all mutations. Accounting for the proportion of aromatics in Yvc1p, this bias is expected to generate 15% aromatic amino acid substitutions among all substitutions, which is slightly higher than the 12% expected from completely random nucleotide substitutions. The probability of encountering six substitutions away from aromatics, one to aromatic, and three not involving aromatics by chance is only 0.00012, even with this bias.

Fig. 1.

The locations of the TRPY1 GOF mutations and their alignment with other TRP channels. (Top) Seven GOF mutations (open stars) with respect to the membrane topology of a TRPY1. From the N terminus, the mutations are F247L, S297W, F380L, Y442C, Y458H, Y473H, and Y473C. (Middle) TM5 and TM6 are redrawn horizontally, marking the positions of F380, Y442, Y458, and Y473 of TRY1. Alignment shows that these aromatic residues are completely conserved among fungal TRPs and are often, although not always, found in animal TRPs. F380 is in the center of a cluster of phenylalanines, and this cluster has its counterparts in TRPY3 of Candida albicans; TPRY4 of C. neoformans, TRPY5 of Aspergills nidulans, TPRY6 from Ustilago maydis, TRPA (painless) of D. melanogaster, TRPM (TRPM2) of Homo sapiens, TRPML of H. sapiens, TRPN (NOMPC homolog) of Danio rerio, TRPP (PKD2) of Mus musculus, TRPV (TRPV1) of H. sapiens, and TRPC (TRP-like) of D. melangaster. (Bottom) Redrawn from Hong et al. (20), showing the alignment of TRPY1's F380 with F550 of Drosophila TRP (asterisk) in a phenylalanine cluster found in TRP channels of Drosophila, C. elegans, and human, as well as in the voltage-gated K⁺ channel Shaker.

Phenotypes in Vivo.

Addition of 1.75 M sorbitol to wild-type yeast cells elicits a relatively small signal, peaking at ≈1.5 × 10⁴ relative luminescence units (RLU) per sec in our test system. Six GOF mutants give responses peaking >2 × 10⁵ RLU (Fig. 2A). Y442C is not as severe as the six, but nonetheless shows a response stronger than that of the wild type (Fig. 2B). When the peak responses to different osmotic challenges are normalized, the wild type has the highest peak at 2.75 M sorbitol (Fig. 2C, black curve). It is not clear why the peak response becomes smaller at an even higher osmotic challenge (3 M). Besides Yvc1, cytoplasmic [Ca²⁺] is governed by the reuptake through exchangers and pumps of the vacuole, as well as other organelles, all influencing the peak responses in vivo. Nonetheless, the maximum peak responses of the GOF mutants are mostly ≈1.75 M instead (Fig. 2C, colored curves). The left shifts of these curves are largely consistent with the GOF's increases in channel open probability (P_o) under most conditions.

Fig. 2.

Osmotic upshock-induced Ca²⁺ release of wild type and the seven GOF mutants. (A) Responses of populations of the wild type and six strong GOF mutant yeast cells to the addition of 1.75 M sorbitol at 30 sec (arrow) as monitored through the luminescence (in RLU) of transgenic aequorin. The wild type (black line) has a small response evident in B. (B) The response to the same osmotic upshock of Y442C, a moderate GOF mutant. Note scale difference between A and B. (C) Peak responses to different sorbitol concentrations normalized to the highest peak observed. Compared with the wild-type responses (black), the strongest peak responses of all seven mutants (colored) are found at lower osmotic upshocks.

The Molecular Functional Defects of F380L.

F380 aligns exactly with F550 of the canonical Drosophila TRP (Fig. 1 Bottom), of which a mutation is known to cause constitutive channel activities and severe retinal degeneration (20). Taking advantage of the yeast preparation, we carried out a detailed single-channel analysis under a patch clamp. Both the wild-type and mutant channels are 320 ps in conductance (in symmetric 180 mM KCl). The wild-type channel rarely opens at 10⁻⁶ M Ca²⁺ (P_o < 0.01), reaches a P_o of ≈0.2 at 10⁻⁵ M Ca²⁺, and approaches a P_o of ≈0.7 at 10⁻³ M Ca²⁺. Although requiring a minimal [Ca²⁺] of 10⁻⁶ M for activation, the F380L channel has P_o values that are significantly higher than those of the wild type at all [Ca²⁺] >10⁻⁶ M. It is almost always open >10⁻⁵ M Ca²⁺ (Fig. 3 A and B). We also examined channel mechanosensitivity. Fig. 3C Upper shows continuous recording from a wild-type patch bathed in two different [Ca²⁺] and subjected to a series of different pressure pulses. Fig. 3C Lower shows the comparable recording from a mutant patch, showing responses to pressure at 2 × 10⁻⁶ M [Ca²⁺]. At 10⁻⁵ M, the ≈100% opening of the mutant channel obscures its mechanosensitivity. From the recordings of five wild-type and five mutant patches that survived the repeated-pressure applications, we are confident that both channels open more frequently upon pressures, although detailed quantitative analyses are beyond our current level of resolution because of unavoidable variations.

Fig. 3.

Single-channel analyses: Wild type versus F380L mutant channel. Recordings carried out in excised cytoplasmic side-out mode. Inward currents are shown downward. (A) Records of currents from a patch with a single conducting unit of the wild type (Left) or the F380L mutant (Right). The patches were perfused with different concentrations of Ca²⁺ as marked. (B) Summary plot of P_o versus log [Ca²⁺] from extended records, portions of which are shown in A (wild type, mean ± SD, n = 6; F380L, n = 4). (C) Channel activation with pressure applied to a single patch with wild-type channels (Upper), continued through two Ca²⁺ concentrations, and one with an F380L channel (Lower). Two [Ca²⁺] are compared with 2 × 10⁻⁶ M Ca²⁺ (Left) and 10⁻⁵ M (Right). Increased responses to increasing pressure pulses are shown, except for the mutant channels at 10⁻⁵ M Ca²⁺ showing maximal activities.

Other GOF Mutations.

None of the GOF mutations alter the unitary conductance or ion selectivity. All of the GOF mutants, except Y442C, have significantly increased P_o values at 10⁻⁵ M Ca²⁺ with no applied pressure (Fig. 4). This finding is consistent with all of the GOF mutations, which have a much stronger luminometric signal than the wild type (Fig. 2). Y458H and Y442C channels also show severe gating kinetics changes. Y458H, located at the cytoplasmic end of TM6, causes prominent flickers with both its major τ_open and τ_closed <10 ms measured at P_o ≈0.5 (18). Y442C at the vacuolar end of TM6 also causes channel flickers, with τ_open and τ_closed <20 ms (Table 1). The other GOF mutants also have kinetic alterations, although not as severe. Like the wild-type channel, all GOF mutant channels can be activated by the addition of Ca²⁺ to the cytoplasmic side or by applied pressure of tens of mmHg. At the present level of investigation, no major changes in either Ca²⁺ sensitivity or mechanosensitivity are detected in these mutants.

Fig. 4.

Phenotypes in channel kinetic of the GOF mutants. (A) Sample trances from the wild type (Upper) and mutants (Lower) in the order of locations from the N- to C-terminal end. Each trace is from an excised cytoplasmic side-out patch apparently containing a single conducting unit. Recordings are conducted at 10⁻⁵ M Ca²⁺ with no applied pipet pressure. (B) A histogram comparing the P_o of the wild type and the GOF mutants at 10⁻⁵ M Ca²⁺ with no applied pipet pressure. Mean ± SD (n = 3).

Table 1.

Kinetic characteristics of the GOF mutants

WT and mutants	Major τopen, ms	Major τclosed, ms
WT	75 ± 18	133 ± 39
F247L	80 ± 22	90 ± 29
S297W	43 ± 12	60 ± 21
F380L	55 ± 15	90 ± 32
Y442C	11 ± 4.7	<20
Y458H*	5.4 ± 3.7	<10
Y473H	54 ± 20	86 ± 34
Y473C	28 ± 13	45 ± 20

Measurements were made at P_o ≈ 0.5, reached by adjusting [Ca²⁺] in each case. Values are given as mean ± SD (n = 3). Shown are the clearly resolvable major mean open times, τ_open, and major mean closed times, τ_closed.

*See ref. 18.

Discussion

We selected mutants with stronger responses to osmotic upshock in vivo (Fig. 2) and found their channels to have altered kinetic (Table 1) and increased P_o values (except Y442C) (Figs. 3 and 4). Surprisingly, the GOF mutations predicted to be in the TM helices all involve aromatic residues, several of which have counterparts in other TRP subfamilies (Fig. 1).

Molecular Activities of Y458H, Y473H, and Y473C Channels.

Y458H exhibits channel currents that flicker rapidly, indicating destabilization of both open and closed states. Among the 18 other Y458 replacements, only the 2 aromatic substitutions maintain wild-type-like kinetics. The findings fit a structural model, in which the aromatic at position 458 near TM6's end acts as an anchor for the channel gate at I455 (18). Here TM6 has an amphipathic helical extension ending near Y473. Both Y473H and Y473C increased P_o values and altered gating kinetics (Table 1 and Fig. 4). Y473 may complement the major anchors at Y458 to control gate movement.

F380L.

By analogy to K⁺ channels (20, 22), the base of TM5 contacts TM6 near the gate. F380L is located at the base of TM5 (Fig. 1 Top) and is near Y458 of TM6 in the homology model (18). As shown in Figs. 3 and 4, the F380L mutation strongly biases the dwell time balance toward the open state. The F380L of the yeast channel and the F550I of the fly channel substitute the phenylalanine with a hydrophobic residue, and both have high P_o. Although F380 is not completely conserved, it is located within a cluster of phenylalanines found in all TRP channels. The coincidence of the yeast and fly GOF phenotypes indicates that this conserved phenylalanine cluster is needed to stabilize the closed conformation of TRP channels in general.

Y442C, S297W, and F247L.

Y442C is a weaker allele in terms of its phenotypes (Fig. 1 Bottom and Table 1). Although located at the opposite end of the gate, Y442 is in TM6, and its distortion may be felt by the entire helix. Interestingly, Tyr-442 has its exact counterparts in TRP-C, −N, −V, and −M, but is replaced by a phenylalanine in TRP-A, −P, and −ML and by a tryptophan in the TRP channel of Crytococcus neoformans (Fig. 1 Top). No nonaromatic substitutions seem to exist. It appears the aromaticity is of structural and functional concern here. By analogy to the voltage-gated KvAP (23) and Kv1.2 (22), TM1 through TM4 may form a separate module surrounding the core. It appears that F247 in TM1 and S297 in TM3 also are needed to maintain the closed–open balance.

GOF Mutations of Ion Channels.

We chose to select for GOF mutations, excluding mutations that subdue or silence the channels because of misfolding or trafficking mishaps. Thus, we focus on subtler changes that cause the channel to tend toward opening. Such rare mutations have been recovered from screens in Escherichia coli for a mechanosensitive channel (24) or in yeast for native (25) and foreign K⁺ channel (26) after single-gene mutagenesis. Among TRP channels, disease or experimental point mutations reported to date all seem to cause reduction or loss of channel activities, including the dominant polycystic kidney disease (PKD) alleles (27) and a dominant-negative construct of TrpV1 (28). The Trp^P365 mutation in the fly is an exception.

Homologous Mutations in the Yeast TRPY1 and the Fly TRPC1 Have Similar Effects.

The founding member of the TRP-channel superfamily was discovered through recessive nonphototactic Drosophila mutants (1, 2, 29). The canonical defect, signified in the channel's official name TRPC, is the premature termination of the photoreceptor potential in the electroretinogram. Unlike other fly Trp alleles, however, is the semidominant Trp^P365, which exhibits sustained receptor potential throughout the light presentation and causes a rapid and massive retinal degeneration (19, 20) because of constitutive activity and, thus, Ca²⁺ (20, 30), similar to the worm's degenrins (15, 16). Note that F380L of TRPY1 is not an engineered mutation based on the knowledge of the F550I mutation of the fly's TRPC1. Both were the results of assumption-free unbiased genetic screens for two different in vivo phenotypes: the fly's phototaxis (19) and the yeast's response to osmotic shock (Fig. 2). The independent origins of these two phenylalanine-to-hydrophobic mutations, both resulting in constitutive activity, indicate the same mechanistic principle operating across TRP subfamilies. Both TRPY1 and TRPC1 are in the center of a cluster of three phenylalanines. Members of this cluster are found in all subfamilies of animal and fungal TRPs (Fig. 1) and possibly even in K⁺ channels (Fig. 1 Bottom) (20).

The Roles of Aromatics in Membrane Protein Structures.

Aromatic residues from different domains or subunits can interact, as in the cuff that surrounds the K⁺ channel filter (31, 32). Some of the aromatics identified here may well be in such interactions. However, their distribution toward the ends of the TM helices suggests another possibility: protein–lipid interaction. Aromatics have both polar and hydrophobic characters and are known to favor the lipid–water interface location (33). Flanking aromatics confer resistance against displacement to experimental peptides embedded in bilayers (34, 35). A survey of residue distribution shows interfacial preference for tryptophan, tyrosine, and histidine, although not for phenylalanine (36). Aromatic belts that contact lipids are found in KirBac1.1 channel (37) and aquaglyceroporin GlpF (38). In our structural model of the core (TM5 through just beyond TM6) of TRPY1, F380, Y458, and Y473 can contact lipids (18). Several of the GOF mutations appear to be at the ends of TM helices (Fig. 1 Top), where the bilayer surface tension is expected to be highest (39). This finding seems to suggest that they may be the force-bearing foci.

Aromatics Apparently Maintain Conformational Balance of TRPY1.

The key finding that the majority of these GOF-gating mutations are aromatics indicates their special role in channel gating: governing the evolutionarily optimized distribution between the open and closed conformations (states). These mutant channels are functional, indicating that the structures in different conformations are largely intact, but become more or less stable. Thus, our results point to the possible role of aromatics in the relative stability of these functional conformations and not in the basic secondary, tertiary, and quaternary structures of the channel. Some of the aromatics pinpointed in the mutations have counterparts in animal TRP homologs (Fig. 1). In some cases, the conservation is in the aromaticity, but not in the identity, of the residues. The role of aromatics in gating may be quite general. In a molecular dynamics simulation, aromatics appear to shift to the interface upon the opening of KirBac1.1 (37). There also are conserved aromatic residues in TRPY1 not pinpointed by the mutations here. We are engineering mutations in these residues to examine and compare the effects.

Prospective.

Overly active mutant TRP channels may be used to further study the basis of excitotoxicity, which injures the nervous system by anoxia, for example. Several of the aromatics identified by this yeast screen have counterparts in other TRP subfamilies. The coincidence of TRPY1's F380L with the fly TRPC's F550I GOF (19, 20) indicates general principles operating among all subtypes of TRP channels. By combining the prowess of yeast genetics with high-resolution patch-clamp analyses of TRPY1, it is hoped that the dissection of this fungal homolog can contribute further to understanding TRP channels in general.

Materials and Methods

Yeast Strains and Media.

The wild-type parental strain, BY4742 (MATα his3Δ leu2Δ lys2Δ ura3Δ), yvc1Δ mutant (a chromosomal KO strain YOR088W; yvc1::km) (11), standard yeast media, yeast extract peptone with glucose (YPD), complete minimal medium with glucose (CMD), and CM with galactose and raffinose (CMGR) were used (40). For aequorin-luminescence measurements, ammonium sulfate in CMD or CMGR was replaced with ammonium chloride (12).

Mutagenesis, Screen, Luminometric, and Patch–Clamp Phenotyping.

These methods are described in ref. 18. Briefly, the plasmid-bearing YVC1 (TRPY1) was passaged through XL1-Red mutator E. coli and mutated at the level of 2–3% loss of prototrophy. No mutational hot spots were observed in the treated, but unselected, plasmids. Although two mutations were encountered at residue 473, sequencing of unselected plasmids indicate that the screen has not been saturated. Plasmid-transformed Δyvc1 yeast clones were individually shocked osmotically, and the Ca²⁺-dependent luminometric signals from the transgenic aequorin were registered with an automated plate reader. Selected clones were rescreened, and their plasmids were retrieved and sequenced. Their ORFs were then excised, subcloned, and eventually inserted into the yeast chromosome by gene replacement before patch-clamp analyses. Preparations of spheroplasts and vacuoles, as well as patch-clamp procedures in excised-patch mode, were as described (13, 18). Results reported here were all from excised cytoplasmic side-out patches bathed in symmetric 150 mM KCl held at −30 mV pipet-positive.

TM Topology Assignment.

Fourteen public domain TM-prediction programs were used to assign the topology of Yvc1p: HMMTOP (Hungarian Academy of Sciences, Budapest, Hungary), DAS (Stockholm University, Stockholm, Sweden), TMHMM (Technical University of Denmark, Copenhagen, Denmark), TMpred (European Molecular Biology Network, Swiss node), SOUSI (Tokyo University of Agriculture and Technology, Tokyo, Japan), Split (University of Split, Croatia), TMAP (Karolinska Institute, Stockholm, Sweden), MEMSAT3 (University of California, San Francisco, CA), PHD (Columbia University, New York), TSEG (Kyoto University, Kyoto, Japan), Phobius (Karolinska Institute) (41), Thumbup (State University of New York, Buffalo, NY), UMDHMM TMHP (University of Maryland, College Park, MD), and TopPred2 (Institut Pasteur, Paris, France) (42). Not all 14 programs assigned the six predicted TM domains canonical of the superfamily, with 14 of 14 predicting TM1, 6 of 14 TM2, 14 of 14 TM3, 12 of 14 TM4, 14 of 14 TM5, and 14 of 14 TM6. There were two distinct conversions on the predicted cytoplasmic end of TM5, with 4 of 14 predicting a more amino-side start. The main 10 were used for the assignment in this case. Several programs also predicted TM domains in the presumptive pore-forming region, as well as a hydrophobic stretch in the carboxyl tail, both of which were rejected based on known superfamily topology. The mean assignments for the ends of TM1–TM6 for those programs, which made assignments, were TM1:AA237–256, TM2:265–285, TM3:299–320, TM4:335–356, TM5:377–395, and TM6:437–459. The SD from these means was <3.5, with the exception of the cytoplasmic end of TM3, which had an SD of 4.5.