Function and phylogeny support the independent evolution of an … – Nature.com

Phylogenetic properties of Deg/ENaC channels from metazoans and non-metazoans

To better understand the relationships of T. adhaerens TadNaC channels to other Deg/ENaC channels, including those from the fellow placozoan Hoiliungia hongkongensis, we used CLuster ANalysis of Sequences (CLANS)27 on a set of 1074 Deg/ENaC channel protein sequences extracted from high-quality gene datasets of representative species spanning the major animal groupings, followed by phylogenetic inference. We tested a range of P value cut-offs for the CLANS analysis (i.e., 1E10, 1E20, 1E30, 1E40, and 1E50), finding in all cases that the sequences formed one major cluster comprised of two inter-connected sub-clusters (Fig.1, Supplementary Data3 to 7). One of these sub-clusters contained the chordate ASIC and BASIC channels, along with the T. adhaerens channels TadNaC1 to 9 and TadNaC11 (and corresponding H. hongkongensis homologs), and the other the chordate ENaC channels with the singleton placozoan homologs TadNaC10 and HhoNaC10. Our analysis is altogether consistent with a previous study19, both also finding the peptide-gated FaNaC and WaNaC channels from lophotrochozoans to associate with the ENaC sub-cluster, and the peptide-gated HyNaC channels from Hydra magnipapillata to associate with the ASIC/BASIC sub-cluster.

Nodes correspond to individual channel sequences and are colored by taxon as indicated by the legend. Edges correspond to BLAST connections with P-values<1E30. The general locations of the chordate ASIC and ENaC channels, the cnidarian HyNaC and NeNaC channels, the lophotrochozoan FaNaC and WaNaC channels, and the C. elegans ACD channels are indicated. Singletons and non-connected clusters with less than five sequences are masked but available in the corresponding CLANS file (Supplementary Data5).

Comparing the various CLANS analyses at different thresholds, we found that decreasing the P value from 1E20 to 1E30 caused numerous sequences to no longer associate with the main cluster, including a large group of ctenophore sequences (Fig.1, Supplementary Data4 and 5). Decreasing it further to 1E40 caused a large group of D. melanogaster PPK channels to no longer associate with the ENaC sub-cluster, and a set of C. elegans ACD channels to lose their relatively strong connectivity with the ASIC sub-cluster (Fig.1), instead forming a single connection with the ENaC sub-cluster (Supplementary Data6). We therefore selected a P value cut-off of 1E30 to isolate a central cluster of sequences for phylogenetic inference, reasoning that this cut-off struck a balance between strategically removing divergent and/or truncated sequences that would interfere with phylogenetic analysis, while being permissive enough to include most PPK channels. In agreement, a previous study that employed a similar CLANS pre-filtering approach prior to phylogenetic analysis but with a P value of 1E50 excluded the D. melanogaster PPK channels19. In our analysis, pre-filtering the sequences at 1E30 resulted in the removal of 200 sequences, which in addition to the noted cluster of ctenophore channels, included numerous singletons and smaller clusters from platyhelminths and cnidarians (Fig.1). Lastly, our clustering analysis revealed that several Deg/ENaC homologs present in the gene data for unicellular eukaryotic species from the clades Heterokonta (i.e., from the SAR supergroup, for Stramenopila, Alveolata, and Rhizaria) and Filasterea, clustered the ASIC and ENaC sub-clusters (Fig.1), corroborating a report that Deg/ENaC channels are present outside of animals, in select unicellular organisms28.

A maximum likelihood phylogeny inferred from the aligned protein sequences, rooted on the Deg/ENaC channel homologs from the unicellular filasterea-related species Tunicaraptor unikontum, reveals strong phylogenetic support for two distinct clades, termed Clades A and B, corresponding to the ASIC and ENaC sub-clusters (Fig.2), which is consistent with another recent phylogenetic analysis8. In both analyses, most TadNaC channels fall within Clade A (TadNaC1 to 9 and 11), forming a sister relationship with chordate BASIC channels. Instead, the singleton channel TadNaC10, along with its orthologue from fellow placozoan Hoilungia hongkongensis, falls within Clade B. Our analysis also identifies several groups of uncharacterized channels that are positioned between the TadNaC and BASIC channels in Clade A, with representatives from chordates (i.e., cephalochordate and urochordate), ambulacrarians (i.e., echinoderm and hemichordate), and lophotrochozoans (i.e., annelid and brachiopod). Our tree also expands the group of C. elegans channels that form a sister relationship with BASIC channels by including the channels ACD-1, ACD-5, and FLR-1, which notably, resemble TadNaC6 and BASIC channels in being inhibited/blocked by external protons13,14,24,29, and ACD-2 which is proton-activated14. Between T. adhaerens and H. hongkongensis, most Clade A Deg/ENaC channel sequences exhibit one-to-one orthology, except for TadNaC4, 6, and 7, for which H. hongkongensis only bears the single homolog, HhoNaC4/6/7. Also consistent with previous reports10,25, ASIC channels within our phylogenetic tree form two distinct subgroups, Groups A and B (not to be confused with Clades A and B), with chordates (vertebrates, urochordates, and cephalochordates) possessing only Group A orthologues, cephalochordates also possessing Group B orthologues, and ambulacrarians and lophotrochozoans only possessing Group B orthologues. Together, these various described channels form a well-supported subclade within Clade A (i.e., subclade I), which is distinct from subclade II which bears representatives from a broad range of bilaterian and non-bilaterian animals. This includes a clade of C. elegans channels bearing ACD-1, ACD-5, and FLR-1, which resemble TadNaC6 and BASIC channels in being inhibited/blocked by external protons13,14,24,29, and ACD-2 which is proton-activated14, and a large clade of arthropod channels including the D. melanogaster PPK channels, of which PPK1 is also proton-activated16. In addition, Clade A subclade II includes two groups of cnidarian channels, one bearing the neuropeptide-gated HyNaC channels from Hydra magnipapillata21 and the proton-activated channel NeNaC2 from Nematostella vectensis8, and several distinct groups of uncharacterized channels from protostomes, ambulacrarians, ctenophores, and poriferans.

The tree was generated with the program IQ-TREE 2 with the best-fit model WAG+F+G4 and rooted with the filasterean-related Deg/ENaC channel homologs. Node support values are for 100 standard bootstrap replicates (green). The asterisks and labels (pink) indicate single channels or clades bearing Deg/ENaC channels that have been characterized as proton-activated.

Clade B similarly subdivides into two major subclades, with subclade I bearing the chordate ENaC channels and the C.elegans mechanosensory channels MEC-4 and MEC-10 and the proton-activated channel ASIC-18,14. Also within subclade I are the placozoan channels TadNaC10 and HhoNaC10, along with a diversity of uncharacterized channels from cephalochordates, ambulacrarians, and protostomes, and a clade of cnidarian channels which includes the N. vectensis proton-activated channel NeNaC148. Clade B subclade II contains a large group of protostome channels which includes the peptide-gated FaNaC and WaNaC channels from annelids and molluscs9, and several uncharacterized representatives from ambulacrarians, cephalochordates, and protostomes (i.e., lophotrochozoans and ecdysozoans including a large clade of channels from Centroides sculpturatus). Lastly, a set of cnidarian channels and Deg/ENaC homologs from the unicellular heterokont Cafeteria roenbergensis form a sister clade relationship with all other Clade B channels.

Altogether, our combined CLANS and phylogenetic analysis provide strong evidence that most TadNaC channels, including the previously described TadNaC6 and the currently described TadNaC2, are phylogenetically closer to BASIC channels than ASIC channels. Furthermore, our analysis corroborates the existence of two major groups of metazoan channels8,19, identifies numerous groups of uncharacterized channels with phylogenetic proximity to channels with known properties, and provides phylogenetic evidence for the existence of Deg/ENaC channels outside of Metazoa.

Previously, we found that the T. adhaerens Deg/ENaC channel TadNaC6 conducts constitutive Na+ leak currents in vitro that are blocked by external protons and Ca2+ ions24 (Fig.3a). Here, we set out to characterize the in vitro properties of a second T. adhaerens Deg/ENaC channel, TadNaC2. Whole-cell patch clamp recording of Chinese Hamster Ovary (CHO)-K1 cells transfected with the TadNaC2 cDNA revealed robust inward macroscopic cation currents elicited by perfusing a pH 5 solution over the recorded cells. No such currents were evident in untransfected cells, but we did observe a small endogenous inward current in these cells that became activated by solutions with a pH of 4 or lower (Fig.3a). For comparison, we also transfected mouse ASIC1a (mASIC1a) which has been extensively studied in vitro, observing robust inward currents at pH 5 with a noticeably faster desensitization than TadNaC2. TadNaC2 whole-cell currents were quite large in amplitude, reaching upwards of 5000 picoamperes (Fig.3b), despite the cDNA not being codon optimized as was required for efficient expression of the cnidarian HyNaC channels in mammalian cells30.

a Sample whole-cell currents recorded for the previously characterized Trichoplax Deg/ENaC sodium leak channel TadNaC6 that is blocked by extracellular protons24, a newly identified endogenous current in CHO-K1 cells that becomes activated upon perfusion of strongly acidic solutions below pH 4.0, and large, prominent proton-activated currents conducted by the in vitro expressed Trichoplax TadNaC2 and the mouse ASIC1a (mASIC1a) channels. b Plot of average peak inward current amplitude (in picoamps or pA) for currents shown in (a) standard deviation. Orange symbols denote values for individual cells/recordings.

Next, we sought to compare the general properties of TadNaC2 and mASIC1a proton-activated currents. Perfusion of external solutions of various pH revealed that TadNaC2 begins activating at pH 5.5, with current kinetics that accelerate from a slow onset non-desensitizing current at pH 5.5, to a faster transient and partially desensitizing current at pH 4.0 (Fig.4a). These currents are markedly different from those of mASIC1a, which began activating at the more basic pH of 6.7, with much faster activation and desensitization evident across all tested values of pH. Notably, the TadNaC2 currents appear biphasic, particularly upon activation with a pH 4.5 solution, with a fast/early transient component followed by a slower/late sustained component. Doseresponse curves generated from these experiments revealed that TadNaC2 is considerably less sensitive to external protons than mASIC1a (Fig.4b), with a pH50 of 5.10.1 vs. 6.70.1, and a Hill coefficient (nH) value of only 1.70.4 vs. 8.42.7. Notably, these values for the mASIC1a channel are closely in line with those reported for the human ASIC1a channel recorded in Xenopus oocytes25,26. Together, the lower pH50 and nH values observed for TadNaC2 indicate a lower binding affinity and reduced cooperativity for extracellular proton binding, more inline with the sensitivity reported for the rat ASIC2a channel31,32,33.

a Sample recordings of TadNaC2 currents (top) and mouse ASIC1a currents (mASIC1a, bottom) elicited by perfusion of solutions with decreasing pH. b pH doseresponse curves for TadNaC2 (n=1417) and mASIC1a (n=713) revealing a right shifted pH50 for the Trichoplax channel relative to mASIC1a, and a smaller Hill coefficient (nH). The values observed for mASIC1a are consistent with previous reports26. c Sample sequential TadNaC2 currents exhibiting rundown or tachyphylaxis similar to mASIC1a. d Plot of average normalized current amplitude standard deviation through successive sweeps for TadNaC2 (i.e., early and late currents at pH 4.5, n=6, and peak current at pH 5.5, n=56) and mASIC1a (peak current at pH 5.5, n=8), revealing decaying amplitudes for all conditions that are statistically indistinguishable from each other (i.e., p>0.05 for one-way ANOVAs comparing raw normalized values for each condition at each pulse). The asterisks indicate statistically significant p values (i.e., <0.05) for pairwise post hoc Tukey tests after one-way ANOVAs of each set of pulses for each condition (TadNaC2 pH 4.5 early: p=1.0E4, F=7.6; TadNaC2 pH 4.5 late: p=2.4E3, F=4.8; TadNaC2 pH 5.5: p=4.3E7, F=14.7; mASIC1 pH 5.5: p=1.4E14, F=38.2). e Sample current recordings for TadNaC2 and mASIC1a before (black traces) and after (red traces) perfusion of 3mM amiloride, revealing a nearly complete block for mASIC1a (at pH 5.5) and only ~50% block for TadNaC2 (pH 4.5). f Plot of average percent block of inward current standard deviation for TadNaC2 (n=8) and mASIC1a (n=7) before and after perfusion of 3mM amiloride. Individual replicates are included as gray circles. B+T indicates the total decay in average current for a successive sweep, which includes the effects of drug block (B) and tachyphylaxis (T), while B indicates the isolated component of drug block alone, obtained by subtracting the average decline in amplitude caused by tachyphylaxis. Denoted p values are for post hoc Tukeys tests after one-way ANOVA (p=1.7E11, F=56.1). g Sample sequential TadNaC2 currents elicited by perfusion of pH 4.5 solutions bearing increasing concentrations of amiloride. h Average amiloride doseresponse curve (n=9) revealing a more pronounced decline in normalized peak inward current with increasing concentration of amiloride, compared to the rundown observed in the absence of drug attributable to tachyphylaxis.

In early experiments, we found that TadNaC2 currents exhibit a non-recovering decay in amplitude upon repeated activation. For example, applying paired 30s pulses of pH 4.5 solution separated by neutral pH wash steps of either 40 or 80s resulted in similar decreases in amplitude of 55.516.5% with a 40s interval vs. 50.749.2% with an 80s interval. Since doubling the interpulse interval from 40 to 80s did not diminish the current decay amplitude, the observed process is not likely due to incomplete recovery from fast/acute desensitization. This feature of TadNaC2 thus resembles the rodent ASIC1a channel which undergoes slow desensitization or tachyphylaxis, a unique process not observed for ASIC2 and ASIC3 proposed to involve a prolonged inactivated state that is distinct from acute desensitization34,35. To better characterize this property of TadNaC2, we employed an experimental paradigm similar to one used previously to study tachyphylaxis of rat ASIC1a in Xenopus oocytes34. Specifically, we applied six 15-second pulses of pH 4.5 or 5.5 solutions over recorded cells expressing TadNaC2 or mouse ASIC1a, separated by 55-second interpulse intervals. Consistent with observations in oocytes, mouse ASIC1a peak currents decayed upon repeated activation at pH 5.5 (Fig.4c), decreasing to 49.49.5% of their original value after 6 pulses (Fig.4d). Similarly, peak TadNaC2 currents declined to 57.920.5% at pH 4.5, and 43.516.4% at pH 5.5, while the late/sustained component of the TadNaC2 current at pH 4.5 declined to 59.522.9%. Analysis of the average data revealed that although the decline in current amplitude for each condition relative to the first pulse was statistically significant, the degree and rate of decline between the different channels and conditions was not.

Next, we tested the sensitivity of TadNaC2 to the general Deg/ENaC channel blocker amiloride, having previously found that the T. adhaerens TadNaC6 channel was potently activated by this drug24, a rare feature also reported for ASIC3 channels36,37. Application of 3mM amiloride almost completely blocked mASIC1, but only partially blocked TadNaC2, altering the current waveform such that the fast early current component was no longer evident (Fig.4e). Given that TadNaC2 and mASIC1 currents, respectively, decay by 19.111.6% and 17.38.3% upon successive activation, we reasoned that a component of the attenuated current amplitude in these experiments was attributable to tachyphylaxis. Subtracting the effect of tachyphylaxis to isolate the amiloride block of both channels reduced the decrease in average peak inward current from 90.34.1% down to 73.04.1% for mASIC1a, and from 60.59.5% to only 41.49.6% for TadNaC2 (Fig.4f). To better characterize amiloride block of TadNaC2, we applied increasing concentrations of the drug while activating at pH 4.5, revealing a continuing decline in peak current amplitude coupled with a marked reduction in the fast transient current with amiloride concentrations greater than 1mM (Fig.4g). Although a component of the decline in current amplitude is likely due to the tachyphylaxis-like property of TadNaC2, there is a marked difference in the current waveforms, in that tachyphylaxis does not markedly alter the macroscopic current waveform (Fig.4c), while high concentrations of amiloride alter the kinetics of the macroscopic current such that the fast transient current is considerably inhibited (Fig.4g). Although these observations suggest that amiloride has a more potent effect on the early compared to the late current, more detailed studies will be required to characterize this phenomenon. Analysis of the decline in peak current as a function of amiloride concentration reveals a more pronounced decay in amplitude compared to tachyphylaxis with amiloride concentrations greater than 100M, with an IC50 of 52.029.6M attributable to the combined effect of amiloride plus tachyphylaxis (Fig.4h).

TadNaC2 resembles mammalian ASIC3 in conducting biphasic macroscopic currents comprised of an early current that activates and desensitizes quickly, followed by a late current that activates and desensitizes more slowly38. These two components of the TadNaC2 current become even more distinguishable at pH 3.5, where two separate peaks can be observed (Fig.5a). We thus wondered whether these two components of the macroscopic current exhibit differences in their ion selectivity. To test this, we employed the bi-ionic reversal potential technique by perfusing different monovalent cations over recorded cells (Li+, Na+, and K+), while maintaining equimolar Na+ in the internal recording solution, and measuring changes in current reversal potential (voltage where currents reverse from inward to outward) when external permeating ions are altered39. This technique allows quantification of permeability ratios of desired cations relative to Na+ (pX+/pNa+, where X+ is the external cation). Recording TadNaC2 currents at different fixed voltages at pH 4.5, with 150mM Na+ on each side of the cell membrane, produced slowly activating currents that lacked a fast transient component (Fig.5b). As expected, these currents reversed from inward (negative) to outward (positive) near zero millivolts (i.e., 0.870.87mV; Fig.5c). Replacement of extracellular Na+ with an equal concentration of Li+, which has a smaller ionic radius than Na+, produced similar currents that reversed near 0mV and lacked a transient component (2.510.72mV), indicating that TadNaC2 is equally permeable to Na+ and Li+. Notably, all our previous recordings made using standard salines with external Na+ and internal K+ or Cs+ ions produced biphasic currents at pH values below 5.5, unlike currents observed under the bi-ionic conditions of Na+In/Na+Out and Na+In/Li+Out. Thus, it appears that the kinetics of the macroscopic current can differ depending on the types of permeating ions present across the cell membrane, an interesting observation that will require deeper analysis in future studies.

a Sample current recordings of the TadNaC2 channel at pH 4 and 3.5, revealing a biphasic current with a fast transient component (i.e., early current), and a slower, sustained (late current) component. The biphasic current becomes more evident at pH 3.5. b Sample proton-activated TadNaC2 currents recorded at different voltages (voltage protocol on top), under bi-ionic conditions of equimolar intracellular Na+ and extracellular Na+ (Na+ ext.) or K+ (K+ ext.). The star and square symbols denote regions of the currents that were measured to determine reversal potentials. c Plot of average reversal potential data ( standard deviation) for the bi-ionic reversal potential experiments, revealing a leftward shift for both the early and late current components of bimodal currents in the presence of external K+ (n=67) compared to external Na+ (n=7) and Li+ (n=4). d Box plot of average reversal potential data, showing statistically significant differences for both the early and late currents when extracellular Na+ was replaced with K+. The denoted p values are from Tukey post hoc tests after one-way ANOVA (p<1E30, F=2569). e Na+/K+ permeability ratios calculated using the bi-ionic reversal potential data, revealing that the late current exhibits higher Na+ selectivity compared to the early current (p value is for a two-sample t-test). f Sample sequential TadNaC2 currents elicited by pH 4.5 solutions bearing increasing Ca2+ concentrations. g Average Ca2+ doseresponse curve (n=12) revealing a similar decline in normalized peak inward current with increasing Ca2+ concentration compared to tachyphylaxis (in constant 2mM Ca2+). h Sample sequential paired currents elicited by pH 4.5 solutions bearing either 0mM or 10mM Ca2+ ions. i Plot of average percent block of peak inward current (e.g., 1 - P2/P1 from (h) 100%) after switching from 0mM Ca2+ to either 0mM or 10mM Ca2+(n=13 and 10, respectively). The denoted p value is for a two-sample T-test.

Instead, replacement of extracellular Na+ with equimolar K+ (i.e., Na+In/K+Out) produced canonical biphasic currents with a fast transient component and a late sustained component (Fig.5b). The occurrence of these two clearly delineated current components allowed us to measure reversal potentials for each, revealing that although both exhibit a negative shift in voltage compared to bi-ionic sodium, the late current exhibited a more marked hyperpolarizing shift compared to the early current (i.e., 60.422.30 vs. 49.011.62mV, respectively; Fig.5c). A box plot of the reversal potential data for the different bi-ionic measurements, coupled with ANOVA and post hoc tests (Fig.5d), substantiates the negative shift in reversal potentials for both the late and early currents in the presence of external K+, reflecting a general preference of TadNaC2 for Na+ over K+ ions. Furthermore, the more pronounced shift in reversal potential for the late vs. the early current indicates that ion selectivity changes over the course of the biphasic current, such that the early current is less selective for Na+ over K+ compared to the late current (pNa+/pK+ permeability ratios of 7.30.5 and 11.01.1, respectively; Fig.5e).

Next, we sought to determine whether external Ca2+ ions can block inward Na+ currents through TadNaC2. Perfusion of a pH 4.5 external solution containing 140mM Na+ and increasing concentrations of Ca2+ revealed a sequential decline in current amplitude (Fig.5f), which however was not statistically different from that attributed to tachyphylaxis (Fig.5g). Nonetheless, 10mM Ca2+ appeared to cause a downward inflection in the doseresponse curve (Fig.5g), suggesting that at this higher concentration, Ca2+ is able to mildly block TadNaC2. We therefore designed a paired pulse experiment aimed at distinguishing the decline in current caused by tachyphylaxis, from that caused by 10mM Ca2+ block. Specifically, we applied paired pulses of pH 4.5 solutions containing either 0mM of 10mM Ca2+ over recorded cells (Fig.5h), and quantified the decline in peak current amplitude of the second pulse relative to the first (Fig.5i). When both pulses lacked external Ca2+, the peak current amplitude declined by 13.24.6%, while addition of 10mM Ca2+ to the second pulse resulted in a decline of 27.65.2%). Thus, 10mM Ca2+ exerts a low-affinity block of the TadNaC2 current of roughly 14.4%.

Deg/ENaC channels like ASIC channels are homo- and/or hetero-trimeric in nature, with each separate subunit forming a ball in hand tertiary structure comprised of wrist, palm, thumb, finger, knuckle, and -ball regions (Fig.6a). Cumulative efforts have uncovered several core molecular determinants for proton activation of ASIC channels, namely a critical histidine residue in the wrist region (H73 in mASIC1), and a lysine in the palm region (K211) situated at the extracellular interface between subunits (Fig.6a)25,40,41,42,43. A protein alignment of several regions bearing these and other determinants for proton-activation of ASIC channels, including the group A ASIC channels from mice (i.e., ASIC1 to 4), selected group A and B channels from Branchiostoma belcheri25, and the singleton group B channel from Lingula anatina10 reveals near complete conservation of the H73 and K211 residues (Fig.6b). The only exception are the proton-insensitive ASIC2b splice variant which lacks H7344, and ASIC4 which is also proton-insensitive and lacks K21145. Indeed, except for the zebrafish ASIC1 homolog zASIC1.110, all functional group A and B ASIC channels that have been experimentally characterized in vitro bear a conserved H73 residue and most bear a K211 equivalent. The mouse ASIC5/BASIC channel, which falls in a separate clade from ASIC channels (Fig.2) and is not activated by protons, lacks both the H73 and K211 residues. These residues are also absent in other Deg/ENaC channels that are sensitive to external protons in vitro, including TadNaC6, the proton-inhibited channel from T. adhaerens24, and the proton-activated channels TadNaC2, human ENaC-6, C. elegans ACD-2, DEL-9, and ASIC-114, D. melanogaster Pickpocket116, and NeNaC2 from the sea anemone Nematostella vectensis8. However, TadNaC2, as well as the mouse ASIC4 and BASIC channels, possess a cationic residue just one amino acid upstream of the K211 position (i.e., R201 in TadNaC2). TadNaC2 and its Hoilungia hongkongensis orthologue HhoNaC2 also possess a conserved lysine one position downstream (K203 in TadNaC2). Furthermore, all of the ASIC channels except for the non-functional ASIC4 isotype bear a conserved aromatic residue 2 positions upstream of H73 (i.e., Y71). Y71 forms an aromatic bridge with a conserved tryptophan (W287) in mouse ASIC1a, shown to be important for coupling conformation changes in the extracellular domain with gating of the pore helices46. This aromatic residue is absent in all included non-ASIC channels except for TadNaC2 and HhoNaC2 which bear phenylalanine and tyrosine residues at this position, respectively (i.e., F70 in TadNaC2), as well as a tryptophan corresponding to W287 (i.e., W276 in TadNaC2). Also notable is that TadNaC2 possesses several protonatable amino acids that are in proximity of the ASIC H73 position, with aspartate and glutamate residues at positions 75 and 77, and a histidine at position 80 that aligns with residues in the palm region placing it proximal to the noted R201 and K203 residues.

a Ribbon diagrams of the chick ASIC1 homotrimeric channel crystal structure (left, PDB number 6VTK), and the AlphaFold-predicted tertiary structure of the mouse ASIC1a subunit (right). The three separate subunits of the homotrimeric channel are colored in red, white, and gray, and the colored circles denote the carbon atoms of critical residues corresponding to the back-colored residues of the mouse ASIC1a channel in the protein alignment shown in (b) (i.e., atoms in blue are within the acid pocket, pink are within the wrist, green are within the palm, and purple are within the finger). The dashed boxes denote structural regions of the single mASIC1a subunit structure bearing these same critical residues. b Protein sequence alignment of the acid pocket (enclosed by blue dashed boxes), wrist (pink), palm (green), and finger (purple) regions of select proton-activated Deg/ENaC channels from cnidarians and bilaterians with TadNaC2, TadNaC6, HhoNaC2,and HhoNaC4/6/7 channels from the placozoans Trichoplax adhaerens and Hoilungia hongkongensis. Residues that are back-colored in black represent conserved residues for proton activation of ASIC channels, while those back-colored red denote key residues that render the ASIC2b splice variant insensitive to external protons. Residues that are back-colored in gray denote protonatable amino acids in TadNaC2 within these key structural regions, some of which are conserved in cnidarian and bilaterian homologs, while those back-colored in brown denote cationic residues in TadNaC2 that flank the critical K211 residue of ASIC channels, also found in several other channels. Notable is the complete conservation of the critical residues H73 and K211 in all included ASIC channels, and their absence in most non-ASIC proton-activated channels including TadNaC2. c Homology model of the homotrimeric TadNaC2 channel structure (left), and AlphaFold-predicted structure of the single subunit, with a similar annotation as described for (a). d Left panels: Close-up view of the acid pocket region of mASIC1a (top) and TadNaC2 (bottom) within corresponding AlphaFold-predicted structures. The six rendered residues in the TadNaC2 channel correspond to residues that align with the six acid pocket residues in mASIC1a as depicted in (a). Right panels: Surface rendering of the acid pocket region of mASIC1a (top) and TadNaC2 (bottom) reveals a stark difference in the electrostatic potential between the two channel subunits. e Close-up view of the wrist and palm regions of mASIC1a and TadNaC2. Apparent in the wrist region is the absence of a critical H73 proton-sensing residue in TadNaC2, but conservation of the aromatic amino acids F70 and W276, which in mASIC1a (i.e., Y71 and W278) form an aromatic bridge critical for channel gating. Instead, TadNaC2 bears a putative proton-sensing amino acid (H80) at the opposite end of a strand that projects from the first transmembrane helix in wrist region (TMH1) to the palm domain, placing it near the residuesR201 and K203 that flank the critical K211 residue of mASIC1a. f Close-up view of the finger and acid pocket regions, with rendered amino acids corresponding to the positions in the ASIC2b splice variant that make the channel insensitive to protons. Also labeled are the equivalent acid pocket residues, and the predicted 1 to 3 helices in the finger region.

Another region associated with proton activation (and desensitization) of ASIC channels is the acid pocket, comprised of a cluster of four acidic amino acids located between the finger and thumb regions of the subunit monomer, and another pair in the palm region close to K211 (Fig.6a, b). In the trimeric channel, the four acidic residues from one subunit and two from an adjacent subunit combine to form a namesake pocket-like structure where protons are thought to bind and affect channel conformation and gating26. Of note, mutation of these glutamate/aspartate residues does not completely disrupt proton activation, and instead, these appear to be more important for channel desensitization26. Accordingly, the group B ASIC channels from B. belcheri and L. anatina lack most glutamate/aspartate residues in the acid pocket (Fig.6b), while they are largely conserved among the group A channels. Furthermore, the two TadNaC channels, as well as the various other non-ASIC Deg/ENaC channels included in the alignment lack most if not all acidic residues at equivalent positions of the acid pocket.

A third region of interest with respect to proton activation is the finger region, where a motif of four amino acids distinguishes the proton-sensitive ASIC2a mRNA splice variant from the insensitive ASIC2b variant (Fig.6b). Specifically, ASIC2a bears a motif of TTN-XXX-H and is proton-activatable, while ASIC2b bears an SKG-XXX-Y motif and is not47. Moreover, introducing the SKG and Y elements of the ASIC2b motif into ASIC2a, together but not separately, abrogates proton activation, and insertion of the finger region of ASIC2a into ASIC1 causes a marked reduction in proton sensitivity making the latter less sensitive to protons similar to theASIC2a channel48. Notably, a histidine residue within the finger region of ASIC2a (H109) is conserved among many of the included Deg/ENaC channels including TadNaC2 (Fig.6b). However, its functional significance remains unclear, with one mutation study reporting no effect on ASIC2a activation42, and a subsequent study reporting a contribution but not a requirement47 In this region, TadNaC2 also bears two protonatable glutamate residues (E104 and E105).

To better infer how the structure of TadNaC2 compares to the well-studied structures of ASIC channels, we generated a homology model of the homotrimeric channel using the crystal structure of chick ASIC1 as a template (Fig.6c; left panel)49. We also predicted the tertiary structures of the monomeric mouse ASIC1 and TadNaC2 channel subunits with AlphaFold (Fig.6a, c; right panels)50. Labeling the carbon atoms of the Y71, W287, H79, K211, acid pocket, and ASIC2 finger motif equivalents in the homotrimeric structure of the chick ASIC1 channel (Fig.6a), and the F70, W276, H80, R201, K203, D413 (single acid pocket residue), and finger motif equivalents in the model of the TadNaC2 homotrimer (Fig.6c), illustrates the general absence of acid pocket residues in TadNaC2. Also evident are the noted differences in the wrist region, where TadNaC2 lacks an H73 equivalent, and in the palm, where the residues H80, R201, and K203 in TadNaC2 are arranged in a triangular cluster at the interface between subunits, in asimilar position as the K211 residue in ASIC channels. Furthermore, the aromatic residues F70 and W276 residues in TadNaC2 are in proximity to each other, suggesting that like Y71 and W287 in ASIC1, these can form hydrophobic interactions.

The predicted structures of the mASIC1 and TadNaC2 monomers also highlight key differences and similarities. First, whereas the cluster of four acid pocket residues in mASIC1 (D237, E238, D345, and D349) are arranged in a tight cluster, the equivalent residues in TadNaC2 are not (S228, P229, L332, and S336) (Fig.6d, left panels). Rendering the electrostatic potential on the surface of the two channel subunits also illustrates a stark difference in the acid pocket region, with the acidic residues of mASIC1a contributing to a highly electronegative surface, while those in TadNaC2 contribute to a surface that is slightly positive and hence unlikely to attract and bind H+ ions (Fig.6d, right panels). In the wrist region, the W276 sidechain at the base of the thumb of TadNaC2 is in a similar position as theW287 sidechainin mASIC1, situatedbetween the Y71 and H73 equivalent residues F70 and threonine 72 (Fig.6e). Furthermore, both the SKN-XXX-H and SEE-XXX-H finger motifs of mASIC1 and TadNaC2 are within a short loop and adjacent descending alpha helix, consistent with the 1 helix identified in the crystal structure of the chick ASIC1a finger region7. However, this helix is predicted to be two helical rotations longer in TadNaC2, with a short loop connecting it to the downstream 2 helix that is also longer than its predicted counterpart in mASIC1 by one rotation (Fig.6f). Lastly, it is notable that the finger regions of the two channels are positioned above the divergent acid pocket, suggesting that any structural alterations taking place in the finger region would be differentially transferred to the thumb and pore regions that mediate channelgating.

Despite lacking key deterministic residues for proton activation, the similar predicted structure of TadNaC2 compared to mASIC1a prompted us to examine whether corresponding structural regions in the placozoan channel bear unique or conserved elements involved in channel gating. Thus, we performed site-directed mutagenesis on selected aromatic or protonatable residues in the wrist region (F70, D75, and E77), protonatable residues in the finger region (E104, E105, and H109), and protonatable or cationic residues in the palm region (H80, R201, and K203) (Fig.6a). To assess changes in H+ sensitivity and gating properties at different pH, we tested each mutant with a series of perfused solutions of various pH to generate doseresponse curves of recorded macroscopic currents (Fig.7ac; plots of individual variants with replicates provided in Supplementary Fig.1).

ac Average pH doseresponse curves standard deviation for wild-type (wt) TadNaC2 and variants bearing amino acid substitutions within the wrist (a), finger (b), and palm (c) regions. d, e Plot of average standard deviation pH50 (d) and normalized peak current amplitude (e) for wild-type (wt) and various point mutated TadNaC2 channels. Red-colored asterisks denote p value thresholds for two-sample t-tests comparing wild-type to mutant values. f Sample whole-cell currents of wild-type TadNaC2 and select mutant variants.

In the wrist region of rat ASIC1a, mutation of the Y71 aromatic residue to a histidine imposes a ~70% reduction in elicited current amplitude, while mutation to alanine completely disrupts activation by protons46. In contrast, analogous mutations of the F70 residue in TadNaC2 had negligible effects on the pH50 (Fig.7d, e), and no effect on average peak inward current amplitude at pH 4.0 compared to the wild-type channel (Fig.7e). Thus, this residue in TadNaC2 most likely does not form a functionally analogous aromatic interaction with the conserved W276 residue in the thumb region, akin to the Y71-W287 interaction in ASIC1 channels. As noted, TadNaC2 bears two protonatable residues in the wrist region (D75 and E77), within a predicted strand that in ASIC channels projects from the H73 residue in the wrist towards the K211 residue in the palm (Fig.6a, b). The E77 residue in TadNaC2 aligns with D78 in ASIC1a and ASIC2a, which when mutated to asparagine in the rat channels disrupts proton activation41,43. In contrast, alanine substitution of E77 in TadNaC2 had no noticeable effect, while mutation of the D75 residue two positions upstream caused a marked increase in the pH50 (Fig.7d, e). Furthermore, neither the D75A nor the E77A mutation caused a change in maximal peak current amplitude (Fig.7f), or in the overall shape of macroscopic currents (Fig.7f). Overall, TadNaC2 appears different from ASIC channels in lacking homologous molecular determinants in the wrist region that areinvolved in proton gating.

In the finger region, single mutations of E104A and E105A had no effect on pH50 or peak current amplitude (Fig.7b, d, e). However, mutation of both together caused a moderate decrease in both metrics, and altered the macroscopic current waveform by diminishing the fast/early component (Fig.7f). A more dramatic effect occurred for the single mutation H109A, which in addition to reducing maximal peak current amplitude (Fig.7e), produced a biphasic macroscopic current and a complete loss of the early current component at pH 4.5 and 4.0 (Fig.7b, f; Supplementary Fig.1h). Interestingly, mutation of the acid pocket residue D345 in mouse ASIC1a (Fig.6b), which is close to the predicted finger region of TadNaC2 where H109 resides (Fig.6f), also imposes biphasic sensitivity to pH, attributed to the loss of one of two separate proton binding sites involved in channel activation (the other being in the palm domain)51. However, the biphasic effect is much more severe for the TadNaC2 H109A mutation, where instead of plateauing between pH 5.0 and 4.0 like mAIC1a, the current amplitude first decreases from pH 5.0 to 4.5, then increases again from pH 4.5 to 4.0 (Fig.7b, f, Supplementary Fig.1h). This atypical feature precluded accurate fitting of the doseresponse data with either standard or biphasic doseresponse curves, since both serve to model strictly incremental processes (i.e., R2 values of 0.64 and 0.68, respectively). Nonetheless, imposing a standard doseresponse curve over the data revealed reduced sensitivity to protons compared to the wild-type channel, with an average pH50 of 4.70.2 vs. 5.20.1 (Fig.7d). Instead, fitting the data with a bimodal doseresponse curve produced a pH50-1 value of 5.50.3 and a pH50-2 value of 4.50.3, both of which are statistically different from wild type (i.e., P values for two-sample t-Tests 0.005 and 0.0005, respectively). However, since the H109A variant shows diminished sensitivity to protons at the threshold pH of 5.5 (Supplementary Fig.1h), this mutant channel is not likely more sensitive to protons at threshold pH values, but rather, has an overestimated the pH50-1 value caused by the poor curve fit. Of note, while macroscopic current amplitudes varied considerably for the wild-type channel, the pH50 values were less variable (Fig.7d, e). Furthermore, we found no correlation between current amplitude and pH50 for the wild-type channel, altogetherindicating that observed differences in pH sensitivity for the H109A mutant and other channels variants was not due to altered current amplitudes. It is alsonotable that the transient current observed at pH 5.0 desensitized more rapidly compared to the wild-type channel, while at more acidic conditions the transient current was absent leaving only a slowly activating sustained current that increased in amplitude from pH 4.5 to 4.0 (Fig.7f). The most severe of all mutations tested was a triple mutation E104A/E105A/H109A, which produced a channel with very weak activation at pH 5.5 and 4.5, completely lacking transient/early currents at all tested pH (Fig.7f). This resulted in the most significant change in proton gating with an average pH50 of 4.40.1 (Fig.7d, e). Altogether, it appears as though the H109 residue, together with E104 and E105, plays an important role in the proton gating of TadNaC2.

In ASIC1a, deletion of the K211 palm residue results in a strong decrease in proton sensitivity, while mutation to glutamate causes a more moderate effect25. In TadNaC2, mutation of the two cationic residues that flank the K211 position, R201 and K203, produced an increase in proton sensitivity with respective pH50 values of 5.30.1 and 5.50.0 (Fig.7c, d). Notably, the R201A mutation also altered the macroscopic current waveform such that the amplitude difference between the early and late components was greater at pH 5.0 and 4.5 compared to wild-type, but not at pH 4.0 (Supplementary Fig.2). Instead, the K203E mutation disrupted the early current such that it only became evident at very acidic pH values (Fig.7f). Deletion of this same residue (K203), to emulate K211 variants of ASIC channels, resulted in an inability to detect currents even with very acidic pH. Alanine substitution of the unique protonatable H80 residue in the palm region, which is proximal to R201 and K203 in our predicted structures (Fig.6), caused the doseresponse data to become more variable, and the pH sensitivity to become biphasic similar to the H109A mutation (Fig.7c, f; Supplementary Fig.1j). Furthermore, and like the K203E mutation, the H80A mutation strongly disrupted the transient current, which was only evident under very acidic pH conditions (Fig.7f). Fitting a standard doseresponse curve over the data revealed a decrease in pH sensitivity compared to wild-type (i.e., pH50=4.70.3; R2 for global fit = 0.91). Instead, a biphasic curve fit produced a pH50-1 value that was not statistically different from wild-type (pH50-1=5.10.7), but a pH50-2 value that was considerably lower (4.30.3; P value for two-sample t-Test 0.0005; R2 for global fit = 0.96). Of note, mutation of a glutamate residue in mouse ASIC1a, just two amino acids upstream of H80 in our protein alignment and in a similar region of the palm domain (Fig.6b, e), also imposed a biphasic sensitivity to pH51. Altogether, these observations indicate that the H80 residue also plays a role in the proton gating of TadNaC2. Furthermore, the R201 and K203 residues also contribute to TadNaC2 gating, however, their mutation did not produce a rightward shift in the pH doseresponse curve as it did for the analogous K211 residue in ASIC channels25, indicating key functional differences. Finally, all tested mutations in the palm region caused a significant decrease in maximal current amplitude (Fig.7e), most extreme for the K203 variant which was either completely non-functional, not trafficking to the cell membrane, or both.

Next, we wanted to determine whether the noted decrease in current amplitude caused by select mutations was due to reduced functionality or a reduction in channel protein expression. Hence, we N-terminally tagged the wild-type channel with enhanced green fluorescent protein (EGFP), as well as the mutants H80A, H109A, and K203 that, respectively, caused moderate, strong, and severe effects on current amplitude. This permitted inference of the total channel protein levels in transfected CHO-K1 cells via EGFP fluorescence quantification, relative to a co-transfected blue fluorescent protein from the empty vector pIRES2-EBFP. Of note, we tested whether tagging the wild-type TadNaC2 channel with EGFP disrupted its function, finding it to conduct proton-activated currents that were visually indistinguishable from the untagged channel (Supplementary Fig.3). Fluorescence micrographs of transfected cells reveal a noticeable decrease in EGFP fluorescence of all three mutant channels relative to wild type (Fig.8a), with respective normalized average integrated density values of 675%, 434%, and 577% for the H80A, H109A, and K203 mutants (Fig.8b). Average integrated density measurements for the co-transfected EBFP were statistically indistinguishable for all transfections, indicating that the differences in EGFP fluorescence were not due to differences in transfection efficiency. Thus, all three of the tested mutations cause a decrease in total protein expression in vitro.

a Representative fluorescence micrographs of CHO-K1 cells co-transfected with pEGFP-TadNaC2 fusion vector (left panels) and an empty pIRES2-EBFP vector (right panels). b Plot of percent average integrated density standard deviation, quantifying the emitted fluorescence of pEGFP-TadNaC2 wild type (wt) and mutant channels, normalized to the average integrated density of wild-type TadNaC2 (n=3 for each transfection condition). EBFP fluorescence was also quantified to determine transfection efficiency. Cyan-colored asterisks denote p value thresholds for Tukey post hoc means comparisons of fluorescence signals between wild-type and mutant channels after one-way ANOVAs (EGFP: p=5.6E11, F=73.6; EBFP: not significant). c Top panel: Western blot of select EGFP-tagged TadNaC2 channel variants in CHO-K1 cell lysates using anti-GFP polyclonal antibodies, comparing total channel protein content (T) with membrane/surface expressed channel protein content (S) for each variant. Bottom panel: Western blot of the lower half of the membrane used in the top panel, using anti-GAPDH (top bands) and anti-EBFP (bottom bands), polyclonal antibodies. d Quantified band intensity (mean gray area) of TadNaC2 bands in (c), relative to the wild type EGFP-TadNaC2 total protein band, revealing decreased total and surface protein expression of TadNaC2 channels bearing mutations, and a near complete absence of membrane expressed variants harboring a K203 deletion, consistent with our inability to record current for this channel in vitro. Bands for each channel variant were also normalized to the intensity of EBFP present in corresponding total protein lanes.

Using a cell surface biotinylation strategy, we also wanted to characterize the effect of mutations on total protein and membrane expressed protein levels in transfected cells. A Western blot probed with anti-EGFP antibodies revealed a marked reduction in both total protein and membrane expressed (surface) protein levels of mutant TadNaC channels relative to wild-type (Fig.8c). Measurements of the mean gray value of the different bands on the blot reveals similar reductions in total protein levels for all three mutants, and notably, extreme reduction of membrane expressed K203 (Fig.8d). Altogether, this data is consistent with our current amplitude measurements and inability to record currents for the K203 variant of TadNaC2.

More here:

Function and phylogeny support the independent evolution of an ... - Nature.com