Tail-anchored protein C-terminal domains stimulate ATP hydrolysis by the P5A-ATPase Spf1p Authors: Elio O Cianci, Luciana R Mazzitelli, Hugo P Adamo Published: 2026 ABSTRACT AbstractP5A-ATPases constitute a distinct branch of the P-type ATPase superfamily implicated in endoplasmic reticulum (ER) quality control through the removal or reorientation of transmembrane segments. Although genetic and structural studies have suggested a role for these enzymes in transmembrane-helix handling, biochemical evidence linking substrate interaction to catalytic activity has remained limited. Here, we show that purified Saccharomyces cerevisiae Spf1p, a prototypical P5A-ATPase, is stimulated by C-terminal domains of tail-anchored (TA) proteins. Fusion constructs containing the transmembrane and C-terminal regions of either Fis1p (FIS) or the bacterial TA protein YgiM (YGIM) increased ATP hydrolysis approximately twofold in a lipid-dependent and vanadate-sensitive manner. Disruption of membrane-proximal basic residues, either by deletion of the FIS C-terminal KKR motif or by alanine substitution in the YGIM C-terminal tail, markedly reduced stimulation, indicating that C-terminal positive charges contribute to activation. YGIM stimulated Spf1p ATPase activity in a concentration-dependent and saturable manner (Km ≈ 0.12 μM) and increased the relative contribution of a high-affinity component in vanadate inhibition. In pulldown assays, binding of YGIM to Spf1p was enhanced when the enzyme was stabilized in an E2P-like conformation using beryllium fluoride. Structural inspection of the E2P state revealed a conserved acidic patch at the luminal entrance of the central cavity, forming a vestibular ring positioned to interact electrostatically with basic C-terminal tails of TA proteins. Altogether, these results support the role of TA proteins as transported substrates of P5A-ATPases and highlight charge-based interactions at the luminal cavity entrance as a key determinant of substrate-dependent activation.Keywords: P5A-ATPase, Spf1p, tail-anchored proteins, ATP hydrolysis, transmembrane helix dislocase, membrane protein quality control, endoplasmic reticulum stress, membrane protein topology, ATP13A1 The P-type ATPase superfamily comprises a large and evolutionarily conserved group of primary active transporters that couple ATP hydrolysis to substrate translocation across biological membranes (1). Based on sequence and functional criteria, P-type ATPases are classified into five major subfamilies (P1–P5). While P1–P3 ATPases transport inorganic cations and P4 ATPases function as phospholipid flippases, the physiological substrates of the P5 subfamily remained elusive for many years. P5 ATPases are subdivided into two clades: P5A-ATPases, represented by Saccharomyces cerevisiae Spf1p and its human ortholog ATP13A1, and P5B ATPases (ATP13A2–ATP13A5), which mediate polyamine transport in vacuolar and lysosomal membranes (2). A major conceptual advance occurred with the demonstration that P5A-ATPases act as transmembrane-helix dislocases in the endoplasmic reticulum (ER), extracting mislocalized or aberrantly oriented membrane segments from the ER membrane (3). This finding shifted the view of P5A-ATPases away from classical small-molecule transporters toward enzymes specialized in membrane protein quality control. In this framework, substrates are not ions or metabolites, but membrane-embedded polypeptides whose topogenesis has failed or requires correction. Genetic and cell biological studies in S. cerevisiae have implicated Spf1p in a wide range of ER-associated processes, including protein glycosylation, ion homeostasis, lipid organization, ER-associated degradation (ERAD), and membrane protein biogenesis (4, 5, 6, 7, 8, 9). Rather than reflecting multiple independent functions, these phenotypes are consistent with a central role for Spf1p in safeguarding the fidelity of membrane protein insertion and topology within the ER. Cells lacking Spf1p accumulate mislocalized membrane proteins, supporting the idea that P5A-ATPases participate in surveillance pathways that recognize and resolve topogenic errors at the level of transmembrane segments. Consistent with this role, P5A-ATPases function as ER-localized quality-control factors that promote correct membrane protein topology or remove proteins that misinsert into the ER membrane (10). Mitochondrial tail-anchored (TA) proteins such as OMP25 and BAK1, which can occasionally enter the ER via the EMC pathway, are selectively recognized and extracted by ATP13A1 (3). In addition, P5A-ATPases prevent topological inversion of signal-anchor type II membrane proteins during biogenesis, with loss of P5A function triggering distinct quality-control responses (11). Based on the structure of Spf1p and genetic analyses of membrane protein topology, it was proposed that P5A-ATPases may employ a negatively charged luminal pocket to recognize positively charged residues exposed by mis-oriented transmembrane helices or luminal loops of membrane proteins, thereby extending their substrate scope beyond tail-anchored proteins (1, 3, 9). This conceptual framework was further supported by recent studies on the human ortholog ATP13A1, which demonstrated charge-based recognition during the biogenesis of multi-spanning membrane proteins (12). However, direct biochemical evidence linking such charge-based recognition to stimulation of the ATPase catalytic cycle had not been experimentally addressed. Despite this functional divergence from classical ion pumps, P5A-ATPases retain the core architecture and catalytic cycle of P-type ATPases and are thought to operate through a modified Post–Albers mechanism (13, 14). In the E1 state, the enzyme binds ATP at the cytosolic N domain and undergoes autophosphorylation at a conserved aspartate residue. Transition to the E2P state is accompanied by large conformational rearrangements that open a central cavity toward the ER lumen (3). Structural studies have revealed that this cavity can accommodate α-helical polypeptides, consistent with a transport cycle adapted for transmembrane-helix handling rather than ion translocation (3, 15). Cryo-electron microscopy structures of P5A-ATPases have further identified distinctive architectural features, including an N-terminal extension and a large cytosolic insertion often referred to as the arm or plug domain (3, 15, 16). In structures of fungal Spf1p, densities corresponding to putative substrate helices have been observed within the central cavity, providing direct structural evidence for polypeptide engagement. Together, these studies support a model in which P5A-ATPases recognize transmembrane segments within a luminally accessible cavity and use ATP-driven conformational changes to reposition or extract them. Even with these advances, the biochemical principles that govern substrate recognition and coupling to ATP hydrolysis remain poorly defined. In particular, direct evidence linking binding of transmembrane polypeptides to stimulation of P5A-ATPase activity has been lacking, largely due to the absence of tractable in vitro systems. To address this gap, we established a minimal biochemical assay using purified S. cerevisiae Spf1p and model substrates derived from tail-anchored proteins. Using ATP hydrolysis as a functional readout, we demonstrate that C-terminal domains of TA proteins stimulate Spf1p activity in a lipid-dependent and vanadate-sensitive manner. Our results provide direct biochemical evidence that substrate-derived polypeptides modulate the catalytic cycle of a P5A- ATPase, establishing a mechanistic link between substrate engagement and ATP hydrolysis in ER membrane protein quality control. Results Selective activation of Spf1p ATPase by the FIS construct containing a C-terminal basic motif P5A-ATPases have been proposed to extract mislocalized mitochondrial TA proteins from the ER membrane (3, 17). Fis1p (mitochondrial Fission one protein) is a TA protein that plays a central role in mitochondrial fission, peroxisome organization, and cellular stress responses. In yeast, Fis1p is mostly present in the mitochondrial membrane (OMM) and in lower amounts (20%) in the peroxisomes (18). Moreover, in C. elegans Fis has been proposed as a substrate of P5A-ATPase-mediated ER quality control, because inactivation of CATP-8 P5A-ATPase causes Fis1 mislocalization and ER fragmentation (19). Using standard molecular biology methods, we generated a model TA substrate by fusing GFP to the C-terminal domain of Fis1p, which contains its transmembrane segment (TMD); this construct was designated FIS. In this configuration, the GFP-tagged N-terminal region is expected to remain cytosolic, whereas the C-terminal region, depending on the orientation adopted upon TMD insertion, would localize either to the ER lumen or to the mitochondrial intermembrane space. This DNA was cloned into the p426 expression vector for heterologous expression in yeast. To verify that our reporter construct behaved as expected, yeast cells expressing FIS were grown in selective medium containing either raffinose or galactose to achieve low and high expression levels, respectively. At low expression levels, fluorescence distribution in wild-type cells localized predominantly to tubular and punctuate structures consistent with mitochondria, whereas in Spf1pΔ cells, fluorescence prominently highlighted a circular ring consistent with the perinuclear ER membrane (Fig. 1). This result agrees with previous reports showing that Spf1p influences the subcellular distribution of Fis1p (20, 21). After induction in galactose, the fluorescence became more diffuse and partially redistributed to the cell periphery and internal membranes in both strains, suggesting partial mistargeting or saturation of the normal insertion pathway. Localization of a GFP–Fis1 tail-anchored reporter in WT and Spf1pΔ cells. Fluorescence microscopy of yeast cells expressing the GFP–Fis1 tail-anchored reporter (FIS) under low (+raf) and high (+gal) expression conditions. WT and Spf1pΔ strains show distinct localization patterns, consistent with the role of Spf1p in TA-protein targeting. Scale bar: 5 μm. The expressed FIS, Spf1, and Spf1p-GFP proteins were purified by Ni-affinity chromatography. Their purity and integrity were evaluated by SDS-PAGE and GFP fluorescence (Figs. S1 and S2). As previously reported, heat denaturation affects the migration behavior of GFP-fused peptides (22). When samples were boiled prior to loading, FIS migrated according to its expected molecular mass of 32 kDa, whereas non-boiled samples retained their native structure and fluorescence and migrated with an apparent molecular mass of ∼25 kDa. The Spf1p-GFP preparations showed the characteristic high-molecular-weight band (>160 kDa) corresponding to the full-length Spf1p protein fused to GFP, together with faint additional bands likely representing minor degradation products or residual contaminants. Fluorescence imaging confirmed the identity of the GFP-tagged species in both preparations. As shown in Figure 2, Spf1p-GFP alone exhibited a basal ATPase activity of 0.52 ± 0.07 U mg–1, consistent with values previously reported by our group and others (23, 24). The FIS construct itself displayed a very low ATPase activity. However, when Spf1p-GFP was incubated with FIS, ATPase activity increased to 0.87 ± 0.15 U mg–1, representing an approximately 1.7-fold enhancement relative to Spf1p-GFP alone. Design of Fis1p -based tail-anchored constructs and their effect on ATPase activity.A, schematic representation of FIS and FISCT constructs used in this study. The fusion includes an N-terminal purification tag and fluorescent reporter followed by the C-terminal region of Fis1p, which contains a single transmembrane helix and a short luminal tail enriched in positively charged residues. B, ATPase activity of Spf1p-GFP in the presence of FIS and FISCT. ATP hydrolysis was measured at 28 °C under standard assay conditions as described under Experimental procedures, in reaction medium containing 50 mM Hepes-K (pH 7.4 at 28 °C), 0.5 mM EGTA, 5 mM NaN3, 5 mM MgCl2, and 3 mM ATP. Reactions contained detergent–lipid micelles (C12E10/asolectin) and Spf1p-GFP at a final concentration of 25 nM, either alone or in the presence of FIS or FISCT. Where indicated, FIS or FISCT was added at a final concentration of 0.5 μM. Bars represent the mean ± SD from four independent biological experiments, each performed in technical duplicate. Individual points represent the mean of the technical duplicates from each independent experiment. Statistical significance was assessed using an unpaired, one-tailed Student’s t test assuming equal variances. Addition of FIS significantly increased Spf1p ATPase activity compared with Spf1p alone (p = 0.004), whereas FISCT did not produce a significant effect (p > 0.05). We also examined the effect of FISCT, a truncated variant lacking the positively charged C-terminal KKR motif. FISCT alone exhibited minimal ATPase activity (0.15 ± 0.08 U mg–1), comparable to that of FIS. When combined with Spf1p-GFP, FISCT produced no significant stimulation (0.57 ± 0.08 U mg–1) indicating that the C-terminal KKR motif is required for this effect. A construct containing the C-terminal sequence of the bacterial TA protein YgiM behaves as Spf1p substrate Having established that Spf1p is selectively stimulated by FIS but not by a truncated variant lacking key residues at the C-terminus, we next examined whether this regulatory interaction could be extended to other TA proteins. To this end, we turned to YgiM, a TA protein from Escherichia coli that has recently been proposed to participate in bacterial–host interactions through mechanisms influencing immune response regulation (25). Previous studies have shown that, when expressed in S. cerevisiae, YgiM localizes to the peroxisomal membrane, whereas in cells lacking Spf1p, it is retained in the endoplasmic reticulum (26). To facilitate its purification and visualization, we constructed the YGIM fusion protein, in which the complete transmembrane and short luminal segments of YgiM is preceded by a His tag and mTFP as soluble fluorescent reporter (Fig. 3). SDS-PAGE analysis of the recombinant fusion indicated a substantially higher purity compared to preparations of FIS (Fig. S2). In addition to its favorable expression profile in E. coli, YGIM provided a valuable model to test whether the substrate-recognition capacity of Spf1p extends to TA proteins of bacterial origin, thereby broadening the relevance of our observations. Design of YgiM-based tail-anchored constructs and their effect on Spf1 ATPase activity.A, schematic representation of the YGIM fusion construct used in this study. The fusion includes an N-terminal purification tag, a TEV cleavage site and a fluorescent reporter followed by the C-terminal region of YgiM, which contains a single transmembrane helix and a luminal tail enriched in positively charged residues. B, ATPase activity of Spf1p-GFP in the presence of YGIM and YGIM-A. ATP hydrolysis was measured at 28 °C under assay conditions identical to those described for Figure 2B. Reactions contained detergent–lipid micelles (C12E10/asolectin) and Spf1p-GFP (25 nM), either alone or in the presence of YGIM or YGIM-A at a final concentration of 0.42 μM. Data represent the mean ± SD from two independent biological experiments performed in technical duplicate. Addition of the purified YGIM to Spf1p-GFP reproducibly increased ATP hydrolysis relative to the basal eluate control (Fig. 3). To assess whether the C-terminal positive charges required for Fis1p-dependent activation are also important for YgiM, we generated a mutant version of the construct in which the luminal RKRK motif was replaced with alanines (YGIM-A). Purified YGIM-A was tested in parallel with the wild-type construct under identical assay conditions. The WT YGIM fusion stimulated Spf1p-GFP activity by ∼1.7-fold, whereas YGIM-A produced only 1.2-fold increase. Thus, although YGIM-A retains a marginal activating capacity, its markedly reduced effect relative to the wild-type peptide indicates that the C-terminal basic residues strongly potentiate activation. Concentration-dependent and saturable activation of Spf1 ATPase by YGIM ATP hydrolysis by Spf1p-GFP was measured in the presence of increasing amounts of the YGIM-containing eluate or the corresponding control eluate lacking the peptide. Only the YGIM-containing samples produced a progressive increase in ATPase activity, whereas the control eluate had no effect (Fig. S3). Basal activity measured was subtracted from each data point to obtain the YGIM-dependent stimulation curve (Fig. 4). The resulting concentration dependence revealed a saturable activation that reached a plateau at peptide concentrations above ∼0.4 μM, with no indication of inhibition at higher levels. For simplicity, the data were fitted by a Michaelis–Menten model with Vmax = 1.4 ± 0.13 U mg–1 and Km = 0.12 ± 0.04 μM. Within the experimental error of the data, other models (Hill, double hyperbola) did not provide a significantly better fit, as indicated by the Akaike Information Criterion (AIC) (27). The saturable activation produced by YGIM suggests that Spf1p engages this peptide through a specific binding interaction rather than a nonspecific effect of detergent or hydrophobic surfaces. Concentration-dependent stimulation of Spf1 ATPase activity by YGIM. ATP hydrolysis by purified Spf1p-GFPwas measured at 28 °C under assay conditions identical to those described for Figure 2B. Reactions contained detergent–lipid micelles (C12E10/asolectin) and Spf1p-GFP (25 nM) in the presence of increasing concentrations of YGIM, as indicated. Basal ATPase activity measured in the absence of YGIM was subtracted from each data point. The solid line represents a fit to a Michaelis–Menten model. Data points represent the mean ± SD of three independent biological experiments performed in technical duplicate. YGIM activates Spf1 ATPase in a lipid-dependent manner that is additive to lipid activation Protein–lipid interactions are essential for the activity of many integral membrane enzymes, including P-type ATPases (28). Consistent with this requirement, ATPase activity of Spf1p-GFPin the absence of added lipids remained very low over the range of enzyme concentrations tested (Fig. S4). To examine the contribution of phospholipids to Spf1p function and whether YGIM-induced stimulation depends on the lipid environment, ATP hydrolysis by purified Spf1p-GFP was measured across a range of lipid concentrations. Asolectin, a natural mixture of phosphatidylcholine, phosphatidylethanolamine, phosphatidylinositol, and other polar lipids, served as the lipid source. ATPase activity was quantified in the absence or presence of a saturating YGIM concentration (0.42 μM) while varying asolectin from 0 to 650 μM. As shown in Figure 5, Spf1p displayed a biphasic lipid-dependence curve: ATPase activity increased sharply with asolectin concentration, reaching a maximum near 200 μM (K0.5 = 52 ± 10 μM), and declined slightly at higher concentrations. A similar decrease in activity was observed at high phospholipids concentration when C12E10 was kept constant, ruling out detergent concentration as the primary cause of lower activity (Fig. S5). Lipid-dependent modulation of basal and YGIM-stimulated Spf1p ATPase activity. ATP hydrolysis by purified Spf1p-GFPwas measured at 28 °C under assay conditions identical to those described for Figure 2B, as a function of asolectin concentration, in the absence (filled circles) or presence (open circles) of YGIM (0.42 μM). Solid lines represent fits of the experimental data using a model incorporating an inhibitory component at high lipid concentrations (continuous lines) or a Michaelis–Menten–type model applied to the low-lipid range (0–220 μM; dashed lines). ATPase activity at 0 μM asolectin was low both in the absence and presence of YGIM. Data points represent the mean ± SD from two independent biological experiments, except for 0 μM asolectin, which corresponds to three independent experiments. Addition of YGIM enhanced activity across the entire lipid range without altering the position of the activity maximum (K0.5 = 43 ± 8 μM). Under these conditions, the maximal rate increased from 1.4 to 2.2 U mg–1, corresponding to an approximately 1.6-fold activation. At 0 μM added asolectin, ATPase activity remained very low and YGIM produced only a minimal effect, indicating that efficient stimulation requires lipid-activated Spf1p. Importantly, YGIM stimulation persisted at saturating lipid concentrations, where basal activity approached a plateau, suggesting that the peptide enhances Spf1p activity beyond lipid-mediated activation rather than simply mimicking the effect of lipids. YGIM promotes a conformation more sensitive to vanadate inhibition Vanadate is a well-established inhibitor of P-type ATPases that acts as a phosphate analog and can trap the enzyme in a transition-state–like configuration associated with phosphoenzyme hydrolysis, thereby blocking progression of the reaction cycle (29, 30). To test whether stimulation of Spf1p ATPase activity by YGIM is sensitive to vanadate inhibition, we measured ATP hydrolysis by purified Spf1p-GFP in the presence of increasing concentrations of vanadate in the absence or presence of YGIM (Fig. 6). As expected, ATPase activity decreased with increasing vanadate concentration. In both conditions, the inhibition curves were better described by a biphasic model (sum of two decreasing hyperbolic terms), revealing high-affinity (μM) and low-affinity (mM) components (Table S1). In the absence of YGIM, the high-affinity component accounted for ∼20% of total activity. Addition of YGIM not only doubled the overall rate of ATP hydrolysis but also increased the relative amplitude of the high-affinity component to ∼37%. These results indicate that YGIM increases the fraction of Spf1p molecules that populate a vanadate-sensitive, dephosphorylation-competent conformation. However, under all conditions tested, the inhibition curves remained biphasic, and a substantial fraction of Spf1p activity exhibited low sensitivity to vanadate despite the presence of YGIM. Vanadate inhibition of basal and YGIM-stimulated Spf1p ATPase activity. ATP hydrolysis by purified Spf1p-GFPwas measured as a function of vanadate concentration, in the absence (filled circles) or presence (open circles) of YGIM (0.42 μM). Solid lines represent fits of the inhibition profiles using a biphasic model comprising two decreasing hyperbolic components corresponding to high-affinity (micromolar) and low-affinity (millimolar) ranges. Data points represent the mean ± SD from two independent biological experiments. Interaction of Spf1p with YGIM is stabilized by the E2P mimic BeFx To determine whether YGIM interacts directly with Spf1p and whether this interaction depends on a specific conformational state of the enzyme, we performed pull-down assays using His-tagged Spf1p immobilized on Ni–NTA resin. Microsomal Spf1p was solubilized in detergent and bound to a 0.5-ml Ni–NTA column in the absence or presence of 200 μM beryllium fluoride (BeFx), a well-established stabilizer of the E2P-like conformation of P-type ATPases (31, 32). Subsequently, 600 μg of purified YGIM lacking the His tag was applied to the column under each condition. After extensive washing with 20 mM imidazole, bound proteins were eluted with 300 mM imidazole. The flow-through (ft), wash (w), and imidazole-eluted fractions (e1, e2) were analyzed by SDS-PAGE followed by Coomassie staining to visualize total protein and fluorescence imaging to detect YGIM. In both experimental conditions, Spf1p was efficiently captured and eluted from the Ni–NTA resin, as shown by Coomassie staining of the eluates (Fig. 7, lanes e1 and e2), indicating comparable loading and recovery of Spf1p with or without BeFx. In contrast, fluorescence analysis revealed a striking difference in the behavior of YGIM. Under BeFx-stabilizing conditions, YGIM co-eluted with Spf1p, producing strong fluorescence signals in the eluates. Without BeFx, fluorescence in the eluates was minimal, whereas the flow-through (lane ft) displayed intense fluorescence, indicating that most YGIM passed through the column without binding. Importantly, no detectable fluorescent signal was observed in the wash fractions, excluding nonspecific retention on the resin. The high fluorescence intensity in the flow-through under BeFx-free conditions confirms that YGIM was present and competent for binding but was not retained, further supporting the specificity of the interaction observed in the presence of BeFx. These results indicate that YGIM binds weakly to Spf1p under basal conditions but much more efficiently when the enzyme is stabilized in the E2P-like state by BeFx. BeFx-dependent stabilization of the interaction between Spf1p and YGIM. Microsomal His-tagged Spf1p was solubilized in detergent, bound to a 0.5-ml Ni–NTA column, and incubated with or without 200 μM BeFx. Purified YGIM lacking the His tag (600 μg) was applied under each condition. After washing with 20 mM imidazole, bound proteins were eluted with 300 mM imidazole. Flow-through (ft), wash (w), and eluted fractions (e1, e2) were analyzed by SDS–PAGE followed by Coomassie staining (upper panel) and fluorescence imaging to detect YGIM (lower panel). A conserved acidic vestibular ring at the luminal entrance of the Spf1p cavity To assess the architecture of the substrate-binding region, we used the E2P cryo-EM structure of Spf1p (PDB 6XMU) and computationally removed the endogenous putative substrate, revealing a well-defined inward-facing cavity (Fig. S6). This cavity is framed by residues located at positions corresponding to canonical ion-binding sites in other P-ATPases, notably conserved residues in M4 (Glu446 and Glu450), M5 (Asn998), and M8 (Gln1113). At the luminally exposed entrance of the cavity, a conserved acidic patch is present in the loop immediately preceding TM8 (residues E1084, E1087, D1091, E1093, and D1095) (Fig. 8A). This loop forms a continuous negatively charged “vestibular ring” around the luminal mouth of the cavity and is conserved across fungal, animal, and metazoan P5A-ATPases (Fig. 8B). Its position and electrostatic potential place this acidic loop at the luminal entrance of the cavity, in close spatial proximity to where C-terminal regions of TA substrates would be expected to reside. Conserved acidic patch in the M7–M8 luminal loop at the entrance of the central cavity of Spf1p ATPase.A, electrostatic surface representation of S. cerevisiae Spf1p in the E2P conformation (PDB 6XMU), shown after removal of the putative endogenous helical peptide. The image depicts a luminal–lateral view highlighting the acidic residues of the M7–M8 luminal loop. Residues E1084, E1087, D1091, E1093, and E1095 cluster to form a continuous negatively charged ring oriented toward the luminal cavity. Numbering corresponds to the Spf1p sequence used in this study. Red and blue indicate negative and positive electrostatic potentials, respectively. B, multiple-sequence alignment of the M7–M8 luminal loop from representative P5A-ATPases and related P5B-ATPases. Sequences correspond to the luminal loop between transmembrane helices M7 and M8 from S. cerevisiae Spf1p (P39986), C. albicans (Q59Q34), C. thermophilum (G0S4Z4), C. elegans CATP-8 (P90747), R. norvegicus Atp13a1 (A6KAB6), M. musculus Atp13a1 (Q9EPE9), and H. sapiens ATP13A1 (Q9HD20). For comparison, the corresponding luminal loop segments from the P5B-ATPases S. cerevisiae Ypk9 (Q12697) and human ATP13A2 (Q9NQ11) are included. Acidic residues (D/E, red) are highlighted. The Length column indicates the number of non-gap residues, and Q represents the calculated net charge of each loop (D/E = −1, K/R = +1). Discussion Purified Spf1p displays a reproducible basal ATPase activity in the absence of defined substrates (7, 13, 16). The molecular origin of this basal activity is not yet fully resolved at the mechanistic level; however, several observations argue that it predominantly originates from Spf1p itself rather than from contaminating ATPases. Moreover, it has previously been shown that the basal ATPase activity measured after expression and purification of a catalytically inactive Spf1p mutant is markedly lower than the basal activity of the wild-type enzyme (13), supporting the conclusion that the activity reported here reflects intrinsic turnover of Spf1p. All kinetic experiments assessing Spf1p ATPase activity were performed with purified Spf1p containing a C-terminal GFP fusion (Spf1p–GFP), whose ATPase activity is indistinguishable from that of non-tagged Spf1p under the assay conditions used. Spf1p-GFPwas purified under conditions previously optimized to minimize copurifying activities (23), and the basal rates observed here are consistent with those reported for highly purified preparations. Nevertheless, we cannot exclude the possibility that differences in basal activity between preparations may partly reflect variations in the amount or composition of endogenous lipids retained during purification, which could modulate Spf1p activity. We therefore interpreted the basal activity as the reference state of the purified enzyme and examined whether TA-derived substrate surrogates could enhance this activity. As model substrates, we tested constructs comprising C-terminal regions of TA proteins that accumulate in the ER in Spf1pΔ cells. This approach was based on the hypothesis that these constructs, which reproduce key structural features of native dislocation substrates, could stimulate Spf1p′s catalytic cycle. Addition of purified C-terminal domains from distinct TA proteins (yeast Fis1p and bacterial YgiM) increased Spf1p-GFP ATP hydrolysis by approximately twofold. The absence of sequence similarity between these activators suggests that Spf1p recognizes substrates primarily through physicochemical features, such as transmembrane helix hydrophobicity and membrane-proximal positive charges, rather than conserved sequence motifs. This broad substrate tolerance is consistent with the proposed function of P5A-ATPases as dislocases that act on diverse mislocalized or aberrant transmembrane helices in the ER. Determinants of TA substrate recognition by Spf1p ER-resident tail-anchored (TA) proteins typically contain short but highly hydrophobic transmembrane (TM) helices and minimal or absent C-terminal basic tails, features that strongly favor stable ER insertion. In contrast, TA proteins that do not normally reside in the ER generally possess intermediate-length TM helices with moderate hydrophobicity together with short, Lys/Arg-rich C-terminal tails (33, 34). Considering these characteristics, both FIS and YGIM closely match the defining signature of non-ER TA proteins. The Fis1p-derived construct (FIS) contains an 18-residue TM helix of low to moderate hydrophobicity followed immediately by a compact, strongly basic tail positioned adjacent to the membrane surface. YGIM contains a longer TM segment (23-residue), slightly more hydrophobic than that of Fis1p, and a more extended, basic C-terminal tail. Despite its greater length, the YGIM TM helix includes a central GGG motif that confers local flexibility, as well as aromatic residues that favor membrane interface partitioning. Together, these examples indicate that Spf1p can recognize TA substrates with some variation in transmembrane helix length and hydrophobicity, consistent with a recognition mechanism that accommodates diverse TA substrates prone to ER mistargeting. Role of C-terminal basic residues in catalytic activation Deletion of the Fis1p C-terminal KKR motif reduced activation to levels that were not significantly different from baseline, indicating that the membrane-proximal basic residues are a key determinant of substrate-induced catalytic enhancement. Importantly, a decrease in activation was observed not only when the C-terminal region was physically truncated, as in FISCT, but also when tail length was preserved, but positive charges were selectively removed, as in YGIM-A. This shared requirement underscores that luminally oriented Lys/Arg residues are functionally necessary regardless of the specific peptide backbone in which they occur. This requirement parallels the charge-based substrate discrimination described for other ER membrane quality control systems, such as the EMC (35). Consistent with our observations, previous in vivo analyses showed that Fis1p tail-anchor mutants lacking the C-terminal positively charged residues remain membrane-associated but lose organelle specificity, redistributing between mitochondria and the endoplasmic reticulum rather than failing to insert into membranes altogether (20). Taken together, these observations support a model in which C-terminal Lys/Arg residues promote productive encounters between TA helices and the Spf1p cavity by stabilizing electrostatic interactions that favor catalytically competent configurations, without being sufficient on their own to define substrate specificity (Fig. 9). Schematic model of the ATP-driven transport cycle of a tail-anchored (TA) substrate by the P5A-ATPase. The diagram illustrates a working model in which ATP binding, phosphorylation, and dephosphorylation drive the conformational transitions that enable capture and inward movement. The TA substrate is shown as transmembrane helix followed by a positively charged C-terminal tail. In the E1 state, the enzyme adopts an open cytosol-facing conformation, with an extended arm/plug domain positioned such that association of the TA substrate is disfavored. ATP binding and phosphorylation generate the E1P state, bringing the membrane-proximal region of the enzyme closer to the TA segment and allowing initial recruitment of the basic tail to the acidic ring. Transition to E2P produces a luminally aligned cavity continuous with the acidic ring, enabling partial insertion of the substrate. Release of inorganic phosphate yields the E2 state, in which the substrate is advanced toward the opposite side of the membrane. Acidic luminal ring preceding TM8 defines the vestibular mouth of the P5A-ATPase cavity Sequence and structural analyses identify an evolutionarily conserved acidic loop immediately upstream of TM8 at the luminal entrance of the Spf1p cavity, defined by the conserved V-D-L-E-x-E/F motif. Although acidic residues are globally tolerated on the non-cytosolic side of transmembrane segments (36), this general topological bias does not readily account for the pronounced and spatially restricted clustering of Asp/Glu residues within the M7–M8 loop of Spf1p. The strict positional conservation and strong negative electrostatic potential of this region support a functional role at the luminal mouth of the cavity. A multistep model for substrate engagement by P5A-ATPases Together, the C-terminal charge requirement of TA substrates and the vestibular ring identified above point to a multistep recognition model in which electrostatic capture at the luminal vestibule precedes deeper engagement of the transmembrane helix within the cavity. In this framework, the acidic vestibule functions as a long-range attractor that increases local substrate residence time and promotes delivery of the hydrophobic segment into a docking-competent trajectory. Such separation between an electrostatically defined vestibular entry site and a deeper hydrophobic docking region resembles organizational principles described for other P-type ATPases handling complex substrates, including P4-ATPase lipid flippases (37). Determining whether the acidic luminal region is required for substrate-dependent activation, and whether it directly contributes to catalytic coupling will require targeted mutagenesis. Lipid dependence and mechanistic implications Spf1p activity exhibited a clear dependence on phospholipids. In the absence of YGIM, ATPase activity increased with asolectin concentration up to an optimal range, followed by a modest decline at higher lipid levels. This biphasic pattern suggests that moderate lipid concentrations favor catalytic activation, whereas excess lipid becomes inhibitory. In the presence of YGIM, ATPase activity was higher across the lipid range but maintained the same biphasic behavior. The failure of YGIM to activate in the complete absence of added lipid indicates that the lipid environment is essential for substrate-dependent stimulation. The combined effects of lipids and YGIM suggest distinct mechanistic roles: lipids likely promote a permissive functional conformation of Spf1p, whereas substrate binding accelerates rate-limiting steps within the catalytic cycle. This dual requirement resembles the lipid–substrate synergy described for ABC transporters (38) and reflects the intrinsic dependence of P5A-ATPase activity on its lipid environment. Vanadate sensitivity reveals conformational heterogeneity underlying basal and substrate-stimulated Spf1p ATPase activity Vanadate inhibition is a classical tool to probe conformational equilibria in P-type ATPases, as it mimics the transition state of phosphoryl transfer and preferentially traps dephosphorylation-competent intermediates (30, 31). In our experiments, Spf1p-GFP exhibited biphasic vanadate inhibition kinetics both in the absence and presence of YGIM, revealing the coexistence of enzyme populations with distinct sensitivities to vanadate. Although vanadate inhibition of P-type ATPases is often reported as monophasic, biphasic inhibition profiles have been documented and interpreted as reflecting conformational heterogeneity within the enzyme population (39). In the absence of YGIM, the majority of ATPase activity (∼90%) corresponded to a low-affinity, millimolar vanadate-sensitive component, indicating that Spf1p predominantly occupies conformations with limited access to transition-state–like intermediates. A similar low-affinity vanadate inhibition has been reported for the purified P4-ATPase Drs2p (40), consistent with the idea that basal ATPase activity can arise from enzyme populations that are catalytically active but poorly coupled to productive transport. As expected for a substrate, addition of YGIM increased total ATPase activity and selectively enhanced the relative contribution of the high-affinity, micromolar vanadate-sensitive component. This shift is consistent with a redistribution of Spf1p toward conformations that more readily access dephosphorylation-competent states en route to an E2-like configuration. Importantly, the persistence of biphasic inhibition under all conditions indicates that YGIM does not eliminate the low-affinity population but instead increases the fraction of enzyme molecules sampling vanadate-sensitive conformations. Because ATPase activity of P5A-ATPases has not yet been measured in their native membrane environment, it remains possible, by analogy to molecular chaperones such as Hsp70, that Spf1p can hydrolyze ATP at a low rate in the absence of client proteins (41). However, the weak sensitivity of basal Spf1p activity to vanadate suggests that this activity reflects an ensemble of conformational states with limited catalytic coupling. Consistent with this interpretation, ATP hydrolysis by purified Spf1p has been shown to correlate with micellar lipid fluidity and detergent composition (25). Importantly, preliminary results from our laboratory indicate that stimulation by TA-derived constructs persists across different detergents, indicating that substrate-dependent activation cannot be attributed solely to detergent-specific effects. Conformational coupling between YGIM binding and catalysis Pull-down assays confirmed the interaction between Spf1p and YGIM, demonstrating that this binding depends on the enzyme’s conformational state. In the absence of BeFx, YGIM associated only weakly with Spf1p, whereas BeFx-stabilized enzyme displayed clear co-elution of the peptide. These findings demonstrate that Spf1p binds YGIM with higher affinity in an E2P-like conformation, consistent with the alternating-access model of P-type ATPases, in which catalytic transitions regulate substrate accessibility. Mechanistic interpretations and limitations Previous ATP13A1 dislocation assays were performed in isolated membrane preparations, where mislocalized TA segments were extracted into the cytosol and required the chaperone SGTA1 to maintain solubility and prevent aggregation (3). In contrast, our experiments were conducted with purified Spf1p reconstituted in mixed detergent–lipid micelles and in the absence of auxiliary factors. Under these conditions, released peptides could in principle partition into empty micelles, although it is also possible that peptide engagement and release occur within the same micellar environment. Within this simplified system, Fis1p- and YgiM-derived constructs, despite marked differences in sequence, hydrophobicity, and length, elicited comparable stimulation of ATPase activity. This convergence suggests that the observed activation may reflect shared physicochemical features rather than complete substrate translocation. One possibility is that enhanced ATP turnover arises from local rearrangements or oscillatory movements of the positively charged C-terminal segments within the micelle, rather than from full extraction of the transmembrane helix. Accordingly, although our data demonstrate that C-terminal domains of TA proteins stimulate Spf1p ATPase activity and support a role in substrate recognition, ATP hydrolysis in this in vitro system remains an indirect readout of substrate engagement. The increase in activity could reflect early recognition events or allosteric stabilization of catalytic intermediates, rather than completion of a full transport cycle. Discriminating between these possibilities will require reconstitution systems that simultaneously monitor ATP hydrolysis and peptide movement, such as proteoliposome-based assays incorporating fluorescent or radiolabeled TA-derived reporters. Finally, P5A-ATPases have been proposed to act on additional classes of membrane proteins beyond TA substrates (3, 12, 15). Whether the substrate features identified here extend to these proteins remains unclear, as the contribution of membrane-proximal basic residues may depend on insertion geometry, transmembrane topology, or local sequence context. Resolving these questions will be essential for defining the full substrate spectrum and physiological roles of P5A-ATPases in ER membrane protein quality control. Experimental procedures Chemicals SDS, yeast synthetic drop-out medium supplement lacking uracil, yeast nitrogen base without amino acids, D-glucose, polyoxyethylene-10-lauryl ether (C12E10), and polyethylene glycol p-isooctylphenyl ether (Triton X-100) were purchased from Sigma-Aldrich. Soybean L-α-phosphatidylcholine (catalog P5638) and vanadium-free disodium ATP were also obtained from Sigma-Aldrich. Complete His-Tag purification resin (Roche). Tryptone and yeast extract were supplied by Difco. PFU DNA polymerase and molecular weight standards were purchased from PB-L and oligonucleotides from Genbiotech (Buenos Aires, Argentina). Unless otherwise indicated, salts and other reagents were of analytical reagent quality. Construction of FIS, FISCT, YGIM, and YGIM-A expression plasmids The DNA coding for FIS was generated by overlap extension PCR. The C-terminal domain of Fis1p (residues 127–155) was amplified from genomic DNA using primers FFIS and RFIS and fused to a fragment encoding N-terminal 9 × His-tagged sgGFP amplified with primers FHIS and RGFP. The resulting 930-bp fragment, containing sgGFP, a GGGAC linker, and Fis1p residues 127 to 155, was cloned into the p426GAL vector by homologous recombination (42). The FISCT construct was generated using the FIS plasmid as template and primers FHIS and REVR151X AAATTGACCTTGAAAATATAAATTTTCCCCTTATCTTAAGAAGAAACTAGCCACGGCCA to produce a truncated version lacking the distal C-terminal region. The PCR product was cloned into p426GAL. DNA constructs encoding YGIM were synthesized and cloned into pET-30a(+) by GenScript. The YGIM construct comprised an N-terminal 9 × His tag, a TEV protease cleavage site, mTFP1, and residues 161 to 206 of the C-terminal region of E. coli YgiM (UniProt P0ADT8). A mutant variant (YGIM-A), in which residues 198 to 201 (RKRK) were replaced by alanine, was generated using the same vector backbone. All constructs were verified by DNA sequencing. Fluorescence microscopy WT BY4742 (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) and Spf1pΔ (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 YEL031w::kanMX4) strains were obtained from EUROSCARF. Yeast cells were transformed with plasmid p426GAL-FIS using the lithium acetate/polyethylene glycol method (43), and transformants were selected on SD–Ura medium (2% glucose, 0.67% yeast nitrogen base, and 0.19% dropout mix lacking uracil). For microscopy, cells were grown overnight in SD, diluted into SD containing 0.1% glucose, and incubated for 8 h at 30 °C. Cultures were then divided and supplemented with raffinose or galactose to a final concentration of 2% (w/v), followed by incubation for an additional 20 h at 30 °C. Cells were collected by brief centrifugation, mounted on glass slides, and imaged using a wide-field fluorescence microscope equipped with a 100 × oil-immersion objective and standard GFP filters. Images were processed and deconvolved using Fiji (44). Expression and purification of Spf1p, FIS, and FISCT in S. cerevisiae Spf1p and Spf1p-GFPwere overexpressed in the S. cerevisiae strain FGY217 (MATa, ura3-52, lys2Δ201, pep4Δ). Purification of Spf1p and Spf1p-GFPwas performed as previously described (23) and is briefly described below. Starter cultures (100 ml) were grown for 24 h at 28 °C with shaking (280 rpm) in SD medium (2% glucose, 0.67% yeast nitrogen base, and 0.19% Ura–dropout mix). Cells were then transferred into 1 L of induction medium (0.1% glucose, 1.25% yeast extract, 0.8% tryptone) and grown for 12 h under the same conditions. Protein expression was induced by the addition of galactose to a final concentration of 2%, and cultures were incubated for an additional 22 h at 28 °C. Cells were harvested by centrifugation and stored at −80 °C until use. Cell lysis was performed by mechanical disruption with glass beads. Frozen pellets were resuspended in three pellet volumes of lysis buffer (50 mM Tris-HCl, pH 7.0 at 4 °C, 300 mM KCl, 500 mM sucrose, 1 mM EDTA, 5 mM NaN3, 1 mM PMSF, 4 mM β-mercaptoethanol). Glass beads were added at a 4:1 bead-to-pellet weight ratio. Cells were lysed on ice by mechanical disruption using a hand-held homogenizer equipped with a whisk attachment, applying 1-min pulses at maximal power, followed by 1-min cooling intervals, for a total processing time of 1 h. Unbroken cells were discarded by centrifugation for 10 min at 3, 000×g, after which the microsomal fraction was collected by centrifugation for 40 min at 80,000×g. The total membrane fraction was resuspended in 10 ml of purification buffer (50 mM MOPS-KOH, pH 7.0 at 4 °C, 10% glycerol, 100 mM KCl, 1 mM MgCl2), homogenized in a glass Potter homogenizer, and stored at −80 °C. Membranes were solubilized at 4 °C for 30 min with Triton X-100 (10% stock) at a ratio of 2 g detergent per g membrane protein. Insoluble material was removed by ultracentrifugation (80,000×g, 40 min), and the supernatant was applied to a Ni-NTA agarose column (2 ml bed volume) equilibrated with 20 column volumes of purification buffer containing 5 mM imidazole, 0.5% C12E10, 1 mM PMSF, and 4 mM β-mercaptoethanol. After sample loading, the column was washed with 20 column volumes of purification buffer supplemented with 25 mM imidazole and 0.05% C12E10. Bound protein was eluted using purification buffer containing 300 mM imidazole and 0.005% C12E10, maintaining 4 mM β-mercaptoethanol at all steps. Fractions containing Spf1p-GFP were identified by fluorescence, whereas Spf1p fractions were detected by the Bradford assay (45). Eluates were pooled, aliquoted, and stored in liquid nitrogen. Imidazole was removed by size-exclusion chromatography. A 3-mL aliquot of purified Spf1p was loaded onto a Sephadex G-50 column (0.5-cm radius × 10-cm height) equilibrated with the same buffer used for Ni–NTA elution, except that imidazole was omitted. Spf1p eluted in the void volume, whereas imidazole was retained by the matrix and recovered in later fractions. Desalting was monitored by intrinsic fluorescence using excitation at 285 nm; Spf1p displayed the expected emission maximum at 330 to 340 nm, while imidazole exhibited a characteristic 370 to 390 nm peak, allowing unambiguous discrimination. The desalted Spf1p sample was concentrated approximately twofold using Vivaspin Turbo 15 ultrafiltration units (30-kDa MWCO) by centrifugation (1062×g, 1 h). FIS and FISCT constructs were expressed and purified following the same procedure described above for Spf1p, with the exception that induction with galactose was carried out for 8 h instead of 22 h. All other steps of cell growth, lysis, membrane preparation, solubilization, and Ni–NTA purification were performed identically. Expression and purification of YGIM and YGIM-A YGIM and the mutant YGIM-A were expressed in E. coli BL21(DE3). A total of 500 ml of LB medium (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl) was inoculated and grown at 37 °C with shaking (280 rpm) to an OD600 of 0.6 to 0.8. Protein expression was induced with 0.5 mM IPTG, and cultures were incubated for 20 h at 28 °C before harvesting by centrifugation. Pellets were stored at −80 °C until processing. Cell disruption was performed by sonication on ice (12 min total; 1-min pulses with 1-min intervals) in lysis buffer containing 150 mM NaCl, 25 mM Tris-HCl pH 8 (at 4 °C), 0.02% NaN3, 10% (v/v) glycerol, 1 mM PMSF, DNase I (25 μg/ml), lysozyme (0.2 mg/ml), and 2.5 mM β-mercaptoethanol. The membrane fraction was solubilized by incubation with Triton X-100 (2 g detergent per gram of total membrane protein) for 30 min at 4 °C. Insoluble material was removed by ultracentrifugation (30 min, 80,000×g), and the supernatant was applied to a 4 ml Ni-NTA agarose column. Bound protein was washed and eluted under standard Ni-NTA purification conditions. Desalting, concentration, and subsequent handling were performed as described for Spf1p. Samples were resolved on 9% SDS–PAGE (46) and visualized using colloidal Coomassie Blue. Band intensity was quantified using bovine serum albumin as a standard. ATPase activity assay All ATPase activity measurements reported in this study were performed using the Spf1p-GFPvariant, in which GFP is fused to the C terminus of Spf1p. This fusion protein exhibits ATPase activity indistinguishable from that of the wild-type enzyme under the assay conditions used (16, 23). ATPase activity was measured essentially as described previously (16), in a reaction medium containing 50 mM Hepes-K (pH 7.4 at 28 °C), 0.5 mM EGTA, 5 mM NaN3, and 5 mM MgCl2 (corresponding to 2 mM free Mg2+) in a final volume of 0.25 ml. Prior to initiating the reaction, a premix was prepared containing Spf1p, either peptide substrate or the corresponding volume of its elution buffer, and lipids. Purified and desalted Spf1p-GFP (15 μl; 1 μg protein) was mixed with 15 μl of substrate or elution buffer and 8 μl of a freshly sonicated lipid–detergent suspension containing 1% (w/v) soybean asolectin and 0.5% (w/v) C12E10. The mixture was preincubated for 8 min at 30 °C and then transferred to the reaction tube. Under these conditions the final concentrations were 420 μM asolectin and 0.016% (w/v) C12E10. After an additional 5 min equilibration, reactions were initiated by adding ATP to a final concentration of 3 mM and incubating for 20 min at 30 °C. ATP hydrolysis was quantified by measuring the release of inorganic phosphate (Pi) using the Baginski method (47). One unit of ATPase activity (1 U) was defined as the hydrolysis of 1 μmol of ATP per minute under the assay conditions. Basal ATPase activity of purified Spf1p (in the absence of peptide substrates) varied among enzyme preparations (0.5–1.3 U mg–1); all comparative experiments were performed using the same enzyme batch. Each data point was measured in duplicate, and assays were run in parallel with a blank containing the complete reaction mixture except for Spf1p. Dependence of ATPase activity on lipids ATPase activity of Spf1p was measured in the presence of increasing concentrations of soybean asolectin (0–650 μM), either in the absence or presence of 0.42 μM YGIM. The molar concentration of asolectin was calculated assuming an average molecular weight of 760 g/mol. Prior to initiating the reaction, a premix was prepared containing Spf1p and either a saturating concentration of YGIM or the corresponding volume of its elution buffer; the required volume of the asolectin/C12E10 suspension was then added to achieve the desired final lipid concentration. ATPase reactions were subsequently carried out as described above. Vanadate inhibition Vanadate stock solution was prepared by dissolving ammonium metavanadate in 30 mM Tris-HCl, pH 7.7, at 25 °C. Increasing concentrations of vanadate were added to a premix containing Spf1p, either substrate or its corresponding elution buffer, and the lipid–detergent mixture, to obtain the indicated final vanadate concentrations in the reaction medium. Pull-down assay YGIM lacking the His-tag was generated by digestion with TEV protease (Addgene plasmid #125194) purified following the Berglund laboratory protocol (48). Proteolysis was carried out overnight at 4 °C in 25 mM phosphate buffer, pH 7.2, containing 125 mM NaCl and 5 mM β-mercaptoethanol, using a 1:10 TEV:substrate molar ratio with gentle agitation. After digestion, the reaction mixture was applied to a Ni–NTA agarose column to remove His-tagged material, and the flow-through containing tag-free YGIM was collected for pull-down assays. Microsomal membranes containing approximately 200 μg of the untagged Spf1p were solubilized and divided into two equal fractions. Each fraction was applied to a Ni–NTA agarose column (2-mL bed volume) and washed with purification buffer supplemented with 25 mM imidazole and 0.05% C12E10. A second wash was performed using the same buffer containing 10 mM MgCl2, and for one column the buffer additionally contained BeFx (0.05 mM BeSO4 and 3 mM NaF). Bound Spf1p was incubated with these wash solutions for 1 h at 8 °C to allow formation of the BeFx-stabilized complex. Equivalent amounts of tag-free YGIM were then added to each column, and the unbound fractions were collected. Columns were subsequently washed with 10 ml of purification buffer containing 20 mM imidazole, with or without BeFx as appropriate. Bound proteins were eluted with purification buffer containing 300 mM imidazole under the same BeFx conditions. Structural modeling and cavity visualization Structural analyses were performed using UCSF ChimeraX (version 1.10.1) (49). The cryo-EM structure of S. cerevisiae Spf1p in an E2P-like conformation (PDB ID: 6XMU) was used as the template for all analyses. Electrostatic surface potentials were visualized using the Coulombic coloring tool. Cavities were identified using the “Find Cavities” module with the following parameters: grid spacing 0.60 Å, inner probe radius 1.40 Å, outer probe radius 4.00 Å, exterior trim distance 2.40 Å and a minimum cavity volume of 5 Å3. Among the detected cavities, the largest lumen-exposed pocket adjacent to TM8 was selected for further analysis. Acidic residues forming the vestibular ring (Glu1084, Glu1087, Asp1091, Glu1093 and Asp1095) were identified based on spatial proximity (<4 Å) to cavity mesh points. Data availability All data supporting the findings reported in this manuscript are included in the article and in the accompanying supporting information. The original raw data underlying the experiments are available from the corresponding author upon reasonable request. Supporting information This article contains supporting information (22). Conflict of interest The authors declare that they have no conflicts of interest with the contents of this article. Acknowledgments We acknowledge the involvement of Paula Grenon in the early stages of this work. We are grateful to Dr Cory Dunn (University of Helsinki) for the generous gift of plasmids related to this study. Author contributions E. C. and L. M. investigation; E. C., H. A., and L. M. validation; E. C., H. A., and L. M. data curation; E. C., H. A., and L. M. formal analysis; H. A., and L. M. methodology; H. A., and L. M. supervision; E. C., H. A., and L. M. writing–original draft; H. A. conceptualization; H. A. visualization; H. A. writing–reviewing and editing; H. A. project administration. Funding and additional information This work was supported by grants from the University of Buenos Aires (UBACyT), with additional support during the initial stages of the project from the National Scientific and Technical Research Council of Argentina (CONICET, PIP 773) and the National Agency for Scientific and Technological Promotion (ANPCyT, PICT 1966). Reviewed by members of the JBC Editorial Board. Edited by Todd R. Graham Supporting information Supporting Figures and Tables References Associated Data Supplementary Materials Supporting Figures and Tables Data Availability Statement All data supporting the findings reported in this manuscript are included in the article and in the accompanying supporting information. The original raw data underlying the experiments are available from the corresponding author upon reasonable request.