Mechanism of 4-aminopyridine inhibition of the lysosomal channel TMEM175

Significance

Transmembrane protein 175 (TMEM175) is a lysosomal cation channel whose activity is critical for lysosomal homeostasis and whose mutation is associated with Parkinson’s disease. The only known pharmacological inhibitor of TMEM175 is 4-aminopyridine (4-AP), a broad potassium channel inhibitor. We determined structures of TMEM175 in the presence of 4-AP and employed molecular dynamics simulations to elucidate how 4-AP inhibits TMEM175 channel activity. Together, these analyses reveal that 4-AP binds in the ion conduction pathway of TMEM175, physically occluding the pore and preventing water and ions from crossing. These studies establish a foundation for developing potent TMEM175 inhibitors that will aid in understanding how lysosomal dysfunction is associated with Parkinson’s disease.

Abstract

Transmembrane protein 175 (TMEM175) is an evolutionarily distinct lysosomal cation channel whose mutation is associated with the development of Parkinson’s disease. Here, we present a cryoelectron microscopy structure and molecular simulations of TMEM175 bound to 4-aminopyridine (4-AP), the only known small-molecule inhibitor of TMEM175 and a broad K⁺ channel inhibitor, as well as a drug approved by the Food and Drug Administration against multiple sclerosis. The structure shows that 4-AP, whose mode of action had not been previously visualized, binds near the center of the ion conduction pathway, in the open state of the channel. Molecular dynamics simulations reveal that this binding site is near the middle of the transmembrane potential gradient, providing a rationale for the voltage-dependent dissociation of 4-AP from TMEM175. Interestingly, bound 4-AP rapidly switches between three predominant binding poses, stabilized by alternate interaction patterns dictated by the twofold symmetry of the channel. Despite this highly dynamic binding mode, bound 4-AP prevents not only ion permeation but also water flow. Together, these studies provide a framework for the rational design of novel small-molecule inhibitors of TMEM175 that might reveal the role of this channel in human lysosomal physiology both in health and disease.

Transmembrane protein 175 (TMEM175) is a cation channel expressed in the lysosome, where it plays critical roles in lysosomal homeostasis (1). Consequently, TMEM175 dysfunction can lead to defects in lysosomal pH regulation, autophagy, and mitophagy (1, 2). Dysregulated lysosomal function is commonly associated with neurodegenerative disease and two variants in the gene encoding TMEM175 have been implicated with Parkinson’s disease (PD). For example, the loss-of-function M393T variant is associated with an increased likelihood to develop PD, while the opposite is observed for the gain-of-function Q65P variant (3–8). Similarly, mice lacking TMEM175 display features consistent with PD, including diminished motor skills and loss of dopaminergic neurons in the substantia nigra pars compacta (7). Despite these strong genetic connections, however, the mechanisms by which TMEM175 dysfunction is associated with the development of PD remain to be revealed. A detailed understanding of the structural and functional characteristics of TMEM175 will be instrumental to that end.

TMEM175 was initially described as a K⁺-selective channel that is evolutionary distinct from other K⁺ channels in amino acid sequence, as well as in its three-dimensional (3D) fold (1, 9–11). In humans, TMEM175 exists as a homodimer with each protomer possessing two repeats of six transmembrane (TM) helices (9). The ion conduction pathway runs along the central homodimer axis and is formed by TM1 and TM7 from each protomer. Near the middle of the ion conduction pore, a ring of highly conserved isoleucine residues (Ile46 and Ile271) forms a narrow hydrophobic constriction site. In the closed conformation of the channel, this isoleucine constriction serves as a physical gate blocking permeation, while in the open conformation it allows permeation of partially dehydrated cations, thereby shaping the ion selectivity of the channel (9, 12). Pore-lining residues near the isoleucine constriction—including Ser45, Thr49, and Thr274—contribute to permeation through transient water-mediated interactions with the permeating ions (9, 12). Recent studies have indicated that in addition to being selective for K⁺ ions over Na⁺ ions, TMEM175 is permeable to protons (13, 14). Moreover, TMEM175 is activated by acidic luminal conditions and has been proposed to regulate lysosomal pH by serving a pH-activated proton channel that effluxes protons into the cytosol in response to excessive luminal acidification (14).

Despite its distinct pore structure and unique selectivity profile, TMEM175 shares some pharmacologic properties with canonical K⁺ channels, including being sensitive to the broad inhibitor 4-aminopyridine (4-AP) (1), an oral medication approved by the Food and Drug Administration to improve walking in multiple sclerosis patients. In neurons, 4-AP inhibits several K⁺ channels, including Kv1.1, Kv1.2, and Kv3.1 (15, 16). Because 4-AP reduces K⁺ efflux from nonmyelinated neurons, thereby increasing the action-potential frequency, this inhibitor has been used to treat demyelinating diseases, such as multiple sclerosis, spinal cord injuries, and encephalopathy (17, 18). Mechanistically, it has been proposed that 4-AP inhibits canonical K⁺ channels by binding to a site near their selectivity filters, hence occluding the ion permeation pathway (15, 19). A mutational analysis of TMEM175 indicated that 4-AP binds within its pore too (11). To date, however, no structural evidence for any of these hypothetical inhibition mechanisms has been obtained, for any K⁺ channel.

Here we present a cryoelectron microscopy (cryo-EM) structure of open-state human TMEM175 in complex with 4-AP at 2.7 Å, in which we observe 4-AP bound in the ion conduction pathway near the isoleucine constriction. Electrophysiological assays and molecular dynamics (MD) simulations reveal the nature of this interaction and confirm the structure captures the blocked state of TMEM175 in lysosomal membranes.

Results

4-AP Is an Inhibitor of TMEM175.

4-AP is the most potent TMEM175 inhibitor that has been identified to date, inhibiting TMEM175-mediated K⁺ currents in enlarged lysosomes with an IC₅₀ of ∼35 µM (1). 4-AP has also been reported to inhibit K⁺ currents from TMEM175 channels in the plasma membrane (11), but this interaction had not been fully characterized. To quantify the effects of 4-AP on TMEM175 channels transiently expressed in the plasma membrane, we measured the dose-dependence of channel inhibition. Using a −80-mV holding potential and a +80-mV pulse, we first examined the effect of increasing concentrations of 4-AP on cation flux by recording currents in the presence of a neutral bath solution containing 150 mM Cs⁺ and a neutral pipette solution containing 75 mM K⁺ and 75 mM Na⁺. From these recordings, we determined that 4-AP inhibits Cs⁺ flux through TMEM175 with an IC₅₀ of 21 ± 3.5 µM (Fig. 1 A and B and SI Appendix, Figs. S1 and S2). We next recorded a voltage family stepping from a holding potential of 0 mV to voltages between −100 mV and +100 mV in 20-mV steps in a symmetrical 150 mM NMDG⁺ condition to examine the effect of 4-AP on proton conduction by TMEM175. Using the −100-mV pulse, we determined that 4-AP inhibits proton flux through TMEM175 with an IC₅₀ of 55 ± 13 µM (Fig. 1 C and D and SI Appendix, Figs. S1 and S2). Thus, 4-AP inhibits both cation and proton flux through TMEM175.

Fig. 1.

4-AP inhibits TMEM175. (A, Left) A schematic of whole-cell patch clamp performed in A and B. (Right) Representative whole-cell electrical recordings of TMEM175-transfected HEK293T cells in the presence of 4-AP. In conditions of 150 mM Cs⁺ (intracellular) and 75 mM K⁺ and 75 mM Na⁺ (extracellular), currents were measured using the following protocol: from a holding potential of −80 mV, the voltage was stepped to +80 mV for 250 ms, then returned to −80 mV for 25 s. SI Appendix, Fig. S1 includes a complete trace. (B) Normalized dose–response curve of three independent experiments of whole-cell patch clamp of TMEM175-transfected HEK293T cells in presence of 4-AP using the +80-mV pulse. Currents are normalized to the trace in the absence of 4-AP. The data were fit to a Hill equation with a Hill coefficient of 1.0 ± 0.3 and IC₅₀ 19 ± 6.6 µM. (C, Left) A schematic of whole-cell patch clamp performed in C and D. (Right) Representative whole-cell electrical recordings of TMEM175-transfected HEK293T cells in the presence of 4-AP. In conditions of 150 mM NMDG⁺ pH 4.7 (intracellular) and 150 mM NMDG⁺ pH 4.2 (extracellular), currents were measured using the following protocol: from a holding potential of 0 mV, the voltage was stepped to voltages between −100 and +100 mV for 120 mS, in 20-mV increments. Currents at −100 mV were used for analysis in D. (D) Normalized dose–response curve of three independent experiments of whole-cell patch clamp of TMEM175-transfected HEK293T cells in presence of 4-AP. Currents were normalized to the trace in the absence of 4-AP. The line through data is the fit to the Hill equation with a Hill coefficient 1.5 ± 0.4 and IC₅₀ 55 ± 13 µM. (E, Left) A schematic of whole-cell patch clamp performed in E and F. (Right) Representative whole-cell electrical recordings of TMEM175-transfected HEK293T cells in the presence of 4-AP, in a symmetrical 150 mM Cs⁺, currents were measured using the same protocol as in C. (F) Normalized current-voltage relationship of 4-AP sensitive Cs⁺ currents. Currents are normalized to the +100-mV pulse. Mean values and SEM from triplicate experiments are shown. (G) Representative whole-cell gap-free electrical recordings of TMEM175-transfected HEK293T cells in the presence and absence of 4-AP. Currents were measured using the following protocol: from a holding potential of 0 mV, the voltage was stepped to either −100 and +100 mV. After a 20-s pulse, the voltage was stepped to either +100 and −100 mV for 20 s, then back to the 0-mV holding potential. The protocol is shown below the recording. (H) Observed kinetic constants of Cs⁺ current upon 4-AP release at +100 mV and −100 mV of membrane potential.

To evaluate the voltage-dependence of inhibition by 4-AP, we recorded currents in the presence and absence of 1 mM 4-AP at different voltages in a symmetrical 150-mM Cs⁺ condition at pH 7.4. By holding the membrane at 0 mV and then stepping to voltages between −100 and +100 mV in 20-mV increments, we found that the 4-AP sensitive current displayed a linear relationship between voltage and current, indicating that in this condition 4-AP blocks TMEM175 equally well at all voltages examined (Fig. 1 E and F and SI Appendix, Fig. S2). To examine the voltage-dependence of 4-AP dissociation, we recorded currents from cells using a symmetrical 150-mM Cs⁺ condition. By stepping to either +100 or −100 mV and transiently perfusing with 1 mM 4-AP for ∼3 s to saturate TMEM175, we could monitor the kinetics of 4-AP dissociation (Fig. 1 G and H and SI Appendix, Fig. S2). At −100 mV, 4-AP quickly dissociated from the channel as the currents were rapidly restored to the baseline levels following perfusing with a bath solution lacking 4-AP. In contrast, at +100 mV, inhibition by 4-AP did not diminish by removing 4-AP from the bath and persisted throughout the remainder of the 20-s pulse. Thus, while saturating concentrations of 4-AP can inhibit TMEM175 activity equally well at all membrane potentials, dissociation of 4-AP is highly voltage-dependent.

To assess the effects of 4-AP on purified TMEM175, we reconstituted TMEM175 into proteoliposomes composed of a 3:1 ratio of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE) and 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1′-rac-glycerol) (POPG) and evaluated the channel activity using a 9-amino-6-chloro-2-methoxyacridine (ACMA)-based proteoliposome flux assay (9). Specifically, K⁺ efflux from proteoliposomes loaded with 300 mM KCl into a solution containing 300 mM NaCl was initiated by the addition of the proton ionophore carbonyl cyanide m-chlorophenylhydrazone (CCCP), resulting in the influx of H⁺ and therefore the quenching of ACMA fluorescence. By evaluating the initial rate of quenching in the presence of varying concentrations of 4-AP, we were able to determine the IC₅₀ is 36 ± 5.6 µM (Fig. 2 and SI Appendix, Fig. S3). Thus, 4-AP inhibits purified, recombinant TMEM175 with a similar potency as TMEM175 expressed in lysosomes.

Fig. 2.

4-AP inhibits purified TMEM175. (A) Schematic of the fluorescence-based proteoliposome flux assay. Purified TMEM175 is reconstituted into lipid vesicles in the presence of 300 mM KCl. To assay, the reconstituted vesicles are diluted into a 300 mM NaCl solution. K⁺ efflux is initiated by the addition of the proton ionophore CCCP, which allows influx of H⁺ to counterbalance the efflux of K⁺. The H⁺ flux is monitored by the fluorescence quenching of H⁺-sensitive dye ACMA. (B) K⁺ efflux from purified TMEM175-reconstituted into liposomes was monitored using a fluorescence-based flux assay. An arrow indicates addition of CCCP to initiate K⁺ flux. Addition of the K⁺ ionophore valinomycin to measure total flux capacity of the liposomes was performed at the end of experiments. Each trace was baseline-corrected by subtracting “no-protein” trace and then fit to a plateau followed by single exponential decay. (C) Normalized dose–response curve of three independent experiments of the fluorescence-based flux assay. All experiments were performed in triplicate and error bars represent SEM.

Structure of 4-AP Bound to TMEM175.

To investigate the mechanism by which 4-AP inhibits TMEM175, we collected cryo-EM images of human TMEM175 vitrified in the presence of 1 mM 4-AP at pH 8.0. Iterative two-dimensional (2D) and 3D classification of the particle images yielded a reconstruction at a resolution of 2.7 Å (SI Appendix, Fig. S1). In contrast to the mixture of open and closed states that were observed for TMEM175 vitrified in the absence of ligand (9, 12), only a single conformation could be reconstructed from the particle images of TMEM175 in the presence of 4-AP, despite multiple attempts to identify alternative conformations using 3D classification and 3D variability algorithms (SI Appendix, Fig. S4 and Table S1). Fitting of the structures of unliganded TMEM175 in the open and closed states into the 4-AP cryo-EM map revealed that the 4-AP complex closely corresponds to the open-state structure, with both TM1 and TM7 adopting kinked conformations (9). Using the open state as an initial model, we built and refined the 4-AP bound structure into the density map.

The resulting model fits well into the density map with good geometry and is nearly identical to the unliganded open-state structure with an RMSD of 0.3 Å for all protein atoms. Like the open-state structure, the pore domain is complete but TM5 and TM6, in the periphery, are poorly resolved. To identify the 4-AP binding site, we subtracted the density map for unliganded TMEM175 in the open state from that of 4-AP–bound TMEM175 (9). This difference map revealed a single strong peak in the ion conduction pathway near the isoleucine constriction (Fig. 3 A–D). Specifically, the peak is on the cytoplasmic side of the constriction and is approximately aligned with Ser45. A triangular-shaped planar density could be resolved at this position in a density-modified and sharpened 4-AP reconstruction (Fig. 3 E and F). However, because TMEM175 is twofold symmetric and 4-AP is small and symmetric, it was not possible to unambiguously model 4-AP into the density peak. It is worth noting, however, that the position of this density exactly coincides with a strong negative peak in the electrostatic potential generated by the protein along the conduction pathway (12), consistent with the fact that 4-AP is positively charged in the physiological pH range. Outside of the 4-AP binding site, which overlaps with the K2 and K3 ion-binding sites (12), the pore is largely unperturbed compared to the unliganded open state. Density peaks are present in the remaining ion-binding sites, including K4 on the luminal side of the isoleucine constriction, and many of the sites of many of the ordered water molecules (SI Appendix, Fig. S5).

Fig. 3.

4-AP binds to the pore of TMEM175. (A and B) Structure of TMEM175 in the presence of 4-AP, shown as top and side views. Difference map calculated by subtracting 4-AP bound map from ligand-free open map is shown as a green surface contoured at 20 σ. (C and D) TMEM175 ion conduction pathway. Side chains for S45, I46, and I271 are shown as sticks. Difference map calculated by subtracting 4-AP bound map from ligand-free open map is shown as a green mesh contoured at 20 σ. TM7 and TM1 are removed for clarity in C and D, respectively. (E and F) TMEM175 ion conduction pathway. Side chains for S45, I46, and I271 are shown as sticks. Ordered K⁺ ions and water molecules are shown as purple and red spheres, respectively. 4-AP bound map is shown as a blue mesh contoured at 12 σ. TM7 and TM1 are removed for clarity in E and F, respectively.

Simulations Indicate Recognition Mode Is Both Dynamic and Specific.

To define the mode of 4-AP recognition and inhibition more precisely, we calculated a series of all-atom MD simulations of TMEM175 embedded in a phospholipid bilayer, in different conditions (Methods and SI Appendix, Fig. S6). Two of these trajectories, each 1-μs long, evaluated the structure of the channel-inhibitor complex described above, with no voltage across the membrane and with 4-AP initially modeled in a tentative configuration compatible with the cryo-EM density map. Throughout these two trajectories the inhibitor remains in the site revealed by the experimental data, but unexpectedly it displays multiple configurations interconverting in the 10- to 100-ns timescale (Fig. 4 A and B and Movie S1). One configuration, however, emerges as the most prevalent in both trajectories, wherein the amino group at the 4-position of the pyridine ring orients toward the cytoplasmic space (Fig. 4 C and D). The second and third most-populated configurations are symmetric with respect to the pore axis and show 4-AP rotated by about 120°, with the amino group oriented instead toward the luminal side (Fig. 4 C and D). Despite being highly dynamic, the plane defined by the pyridine ring relative to the protein is constant throughout the simulations. That is, 4-AP does not rotate about the pore axis, once bound.

Fig. 4.

Coexisting binding poses for 4-AP in the pore of TMEM175. (A) Time traces of the position of 4-AP along the TMEM175 pore axis in two independent MD simulations (black traces), compared with the average position of the Cβ atoms of I46 and I271 (red traces). At the beginning of the trajectories, the position of the inhibitor coincides with the density signal observed in the cryo-EM map. (B) Orientation of 4-AP while bound to TMEM175, defined by the angle between the pore axis (from cytosolic to luminal side) and the vector that connects the N atoms in the amine and pyridine groups (see Inset of C). (C) Probability distribution for the same descriptor of the 4-AP orientation, combining the samples in both MD trajectories. (D) Close-up of the inhibition site, depicting the three principal binding poses in the probability distribution shown in C. Only selected elements of the channel structure and the simulation system are shown, for clarity. The figures depict one representative snapshot of 4-AP (and the protein) for each pose; a density map reflecting the complete configurational ensemble for 4-AP in each case is shown as a blue mesh. (E–G) Characteristic interaction patterns stabilizing each 4-AP binding pose, represented as probability distributions as function of the orientation of the inhibitor relative to the pore axis. E quantifies the total number of contacts between the CH groups in the 4-AP pyridine ring and the hydroxyl groups in Ser45 in each of the two flanking TM1 helices (Methods). F and G quantify the hydrogen bonding contacts water molecules and the pyridine and amine N atoms, respectively. Lastly, H quantifies the number of contacts between the hydroxyl groups in Ser45 and the amine group in 4-AP. See also Movie S1.

Further examination indicates this intriguing conformational specificity is dictated by Ser45, which line the pore in each of the TM1 helices. Specifically, we observe that the hydroxyl oxygens of these two sidechains, which are electronegative, coordinate the electropositive outer edge of the pyridine ring (SI Appendix, Fig. S4). Indeed, in the primary binding pose the pyridine ring forms up to four concurrent CH⋯O contacts with Ser45 (Fig. 4E). In addition, this pose is stabilized by multiple H-bonds with water, principally via the amine group but also the pyridine ring (Fig. 4 F and G). In the secondary binding poses, this network of polar contacts is reconfigured, and it is now the amine group that forms a strong NH⋯O with one of the Ser45 sidechains, while the pyridine ring retains some of its CH⋯O contacts with the other (Fig. 4 E and G). The overall hydration of 4-AP in these poses is diminished, however, which likely explains why they are less populated in simulation. In summary, our results indicate that once 4-AP reaches its binding site near the isoleucine constriction, it is driven to become coplanar with the two hydroxyl sidechains of Ser45, as they complement the strong dipole and quadrupole of the amine and pyridine groups, respectively; this configuration also permits this cationic inhibitor to remain hydrated. It is worth noting that Ser45 is not conserved in the equivalent position in TM7, unlike Ile46; instead, it is substituted by Ala270, explaining why 4-AP does not rotate around the pore axis during our simulations. Similarly, Ser45 is not conserved in the prokaryotic TMEM175 homologs, which are insensitive to 4-AP, and has been previously suggested to at least partially explain their insensitivity to 4-AP (11).

Simulations Indicate Inhibition Mechanism Is Both Steric and Electrostatic.

To examine the mechanism by which 4-AP inhibits channel function, we compared the two trajectories described above for the 4-AP–TMEM175 complex with an additional trajectory of TMEM175 in the unliganded, open state, again with no voltage across the membrane. This comparison indicates that 4-AP inhibits the channel by precluding both K⁺ access to the center of the pore, as well as water flow across it (Fig. 5). As reported in our previous study of the conductive properties of open-state TMEM175 under a TM voltage, K⁺ ions encounter a steep, rate-limiting free-energy barrier at the isoleucine constriction, but the vicinity of this constriction can be accessed with relative ease, as the features of the free-energy landscape there are much shallower (12). The trajectory generated here for the unliganded channel without an applied voltage, but identical KCl concentration conditions, reaffirms that observation (Fig. 5 A and B). The positive charge carried by 4-AP, however, pushes the boundary of this easily accessible region away from the isoleucine constriction. Specifically, on the cytoplasmic side of the pore, where 4-AP binds, sites K1 and K2 are entirely eliminated and site K3 is severely inhibited; on the luminal side, across the constriction, site K4 is eliminated too, while the occupancy of the K5 site is diminished fivefold (Fig. 5 A and B). The exclusion of K⁺ ions from the pore implies 4-AP dissociation is unlikely to result from direct competition with K⁺ ions, once the inhibitor reaches its binding site (see next section). In addition to this electrostatic effect, 4-AP also blocks the pore sterically. Our simulation of the unliganded open state shows TMEM175 is clearly permeable to water; specifically, we observe ∼300 permeation events, in one or other direction, in the span of 1 μs (Fig. 5C and Movie S2). In contrast, in the presence of 4-AP the number of permeations in the same timeframe is nearly zero (Fig. 5C and Movie S1). As mentioned, 4-AP is hydrated while bound, and highly dynamic; however, its pyridine ring, packed against the isoleucine constriction, seals off the conduction pore completely, irrespective of the binding pose. Thus, 4-AP blocks ion permeation by inhibiting K⁺ (or Na⁺) binding to transient binding sites near the isoleucine constriction. In addition, by excluding water molecules from this constriction, 4-AP precludes the formation of a continuous water wire through the TMEM175 pore, therefore blocking proton permeation.

Fig. 5.

4-AP precludes K⁺ permeation and water flux across TMEM175. (A and B) Accessibility of the interior of the TMEM175 pore to K⁺ ions, for both the apo channel (blue lines) and the channel bound to 4-AP (red lines), based on unbiased MD trajectories of 1 μs with 100 mM KCl and at zero TM voltage. Accessibility is defined here as the mean K⁺ concentration at a given distance from the center of the isoleucine constriction, relative to the bulk value. Data are shown separately for (A) the cytosolic and (B) the luminal sides of the pore. Note that while no K⁺ ions were expected to traverse the constriction in these simulation conditions, access to all binding sites (K1 through K5) within the pore was expected to be largely unhindered in the apo state (9), as was indeed observed (blue lines). (C) Flux of water molecules across TMEM175 pore, measured at the isoleucine construction, in simulations of the apo channel (blue lines) and of the channel bound to 4-AP (red/orange/gray/black). Water fluxes toward the luminal side and the cytosolic side are quantified separately, for each of the 1-μs MD trajectories calculated. See also Movie S2.

Simulation Confirms Consistency between Electrophysiology and Structural Data.

To further ascertain that the binding site revealed by the cryo-EM density map corresponds to the mode of 4-AP inhibition probed by our electrophysiological assays, we used MD simulations to evaluate whether 4-AP dissociation from TMEM175 is indeed voltage-dependent, as observed experimentally. To that end, we calculated four independent MD trajectories with 500 mV applied across the membrane (negative on the cytoplasmic side) and monitored the position of 4-AP and K⁺ ions within the channel pore, while also evaluating the time-averaged electrostatic potential across the simulation system, relative to the no-voltage condition (Methods). As shown in Fig. 6A, all simulations revealed full or partial dissociation of 4-AP within 250 ns, in stark contrast to the stability observed at 0 mV in the microsecond timescale. Further examination of the calculated trajectories reveals that 4-AP dissociation is not induced by the occasional entry of luminal K⁺ into the pore (accessing the K5 site), through a knock-on mechanism, as K⁺ ions cannot access the subsequent binding site nearest to the isoleucine constriction (K4) while 4-AP is bound. Instead, the simulated trajectories suggest that 4-AP dissociation is fostered by the gradient in the TM potential, which imposes a preferred directionality on the spontaneous displacements of 4-AP within the pore, favoring those directed toward the cytoplasm. Indeed, a quantification of the electrostatic potential across the simulation system (Fig. 6B), based on the MD trajectories, clearly shows that the binding site revealed by the cryo-EM structure is deep inside the potential gradient; that is, the dynamics of 4-AP at and near this site is strongly influenced by an electric field, in this case directed toward the cytoplasmic end of the pore. These simulation results corroborate the consistency between our electrophysiological and structural data.

Fig. 6.

Inhibition site implies voltage-dependent dissociation. (A) Time traces of the position of 4-AP along the TMEM175 pore axis in four independent MD simulations (black traces) wherein a voltage of 500 mV is applied across the membrane (positive on the luminal side), compared with the average position of the Cβ atoms in I46 and I271 (red traces). Traces for K⁺ ions that reach the entrances of the pore are also shown (gray traces), highlighting those that enter the pore (blue traces). (B) Electrostatic potential resulting from the application of an extrinsic transmembrane voltage (of 500 mV), along the direction parallel to the pore axis. The locations of the 4-AP binding site, of the isoleucine constriction, and of two sidechains marking the entrances to the pore on the cytosolic and luminal sides, are indicated.

Discussion

4-AP has long been used as a probe to study the mechanism of K⁺ channels, and it is also used to treat multiple sclerosis patients. While no structures exist of this inhibitor bound to any canonical K⁺ channel, several lines of evidence led to a model where 4-AP binds to a site near the selectivity filter (19). There, 4-AP is predicted to inhibit K⁺ channels by physically occluding the pore and preventing the flow of ions, as well as by favoring adopting of a closed channel state. Our structural, electrophysiological, and computational analyses show that despite its complete lack of homology with canonical K⁺ channels, TMEM175 too recognizes 4-AP near its selectivity filter, the isoleucine constriction, resulting in a nonconductive state that is impermeable to both cations and protons. Moreover, as with voltage-gated K⁺ channels (19,20), 4-AP appears to inhibit TMEM175 in a state-dependent manner. Specifically, our cryo-EM analysis indicates that 4-AP is recognized only by the open state. Although 4-AP is highly dynamic when bound to TMEM175, neither of the predominant binding poses can be sterically accommodated in the structure of the closed pore, explaining why 4-AP is biased toward to the open state. For canonical K⁺ channels, the state dependence of 4-AP has been proposed to arise from the inaccessibility of the biding site in the closed state, rather than a steric interaction that would prevent channel gating. For example, structures of rat Kv1.2, human Kv1.3, and human Kv3.1 all possess large vestibules, and a selectivity filter that would be sufficient to accommodate 4-AP (SI Appendix, Fig. S8) (23–25). Moreover, similar to TMEM175, a K⁺ ion occupies the putative 4-AP binding site in the structure of rat Kv1.2 (24). Thus, despite its unique structural and functional characteristics, the cryo-EM structure of TMEM175 bound to 4-AP reveals the molecular basis for mechanistic features shared with canonical K⁺ channels.

4-AP inhibits TMEM175 with an apparent affinity of 20 to 40 µM in lysosomes, in the plasma membrane and when reconstituted into lipid vesicles. However, this apparent affinity masks the dynamic nature of 4-AP binding. In simulation, we observed that 4-AP predominantly adopts three binding poses, stabilized by alternate interaction patterns, but relatively short-lived. However, unlike most ligands, 4-AP does not need to dissociate from the channel to interconvert among binding poses. Indeed, it remains at its binding site, continuously preventing ion and water flow through the pore, even while transitioning from one pose to another. Thus, while each pose is quite short-lived, the rapid interconversion between poses implies that 4-AP preserves much of its rotational entropy in the bound state, and therefore, it has a significantly greater affinity for TMEM175 than would be possible for any of the individual poses.

Investigations into the physiological and pathophysiological roles of TMEM175 have been to date limited to genetic perturbations (1, 2, 5). To further our current understanding, development of novel chemical probes will be necessary. While 4-AP is useful for characterizing the electrophysiological properties of TMEM175 in isolation, its lack of specificity against canonical K⁺ channels limits its value for studies based on cells and organisms. Given the unique molecular architecture of TMEM175, however, inhibitors and probes that are more selective and yet sufficiently potent should be within reach. These probes will aid in better understanding the physiological roles of TMEM175 and in particular the effects of its unique ion selectivity profile, being ∼10,000-fold selective for protons over K⁺ ions. In the lysosome, due to the greater than 5,000-fold greater concentration of K⁺ in the cytosol compared protons in a fully acidified lysosome, it likely permeates both protons and K⁺ ions and it will be critical to understand how these fluxes influence lysosomal pH. We posit that the structure of the complex with 4-AP presented here will provide a template for the rational design of such probes and is therefore a stepping stone toward deciphering the role of TMEM175 in lysosomal homeostasis and PD.

Methods

Electrophysiological Analysis.

Electrophysiological recordings of TMEM175 constructs were performed in HEK293T cells (ATCC CRL-3216). HEK293T cells were cultured in DMEM supplemented with 10% FBS. To transfer cells in single dishes, cells were detached by trypsin treatment. The detached cells were transferred to poly-Lys–treated 35-mm single dishes (FluoroDish, World Precision Instruments) and incubated overnight at 37 °C in fresh media. Cells in a single dish were transfected with 1.25 μg of c-term EGFP tagged hTMEM175 plasmid using 3.75 μg of PEI 25 k (Polysciences). Electrophysiological recordings were performed 48 to 72 h after transfection. Prior to recording, media was replaced with a bath solution containing 145 mM Na-methanesulfonate (MS), 5 mM NaCl, 1 mM MgCl₂, 1 mM CaCl₂, 10 mM Hepes/Tris pH 7.4, or 72.5 mM K-MS, 72.5 mM Na-MS, 2.5 mM KCl, 2.5 mM NaCl, 1 mM MgCl₂, 1 mM CaCl₂, 10 mM Hepes/Tris pH 7.4. Ten-centimeter-long borosilicate glasses were pulled and fire polished (Sutter Instrument). The resistance of a glass pipette filled with a pipette solution containing 150 mM Cs-MS, 5 mM MgCl₂, 10 mM EGTA, 10 mM Hepes/Tris pH 7.4 were 3 ∼10 MΩ, and GΩ seals were formed after gentle suction on the cell membrane. All recordings were performed in whole-cell patch-clamp configuration. The dose-dependence of 4-AP inhibition was examined by stepping from a holding potential of −80 mV to +80 mV for 250 ms, followed by a return to −80 mV in the presence of varying concentrations of 4-AP. For electrophysiological recordings in Fig. 1 C and D, prior to recording, media were replaced with a bath solution containing 145 mM NMDG-MS, 5 mM NMDG Cl, 1 mM MgCl₂, 1 mM CaCl₂, 10 mM citrate/Tris pH 7.4. The glass pipettes were filled with a pipette solution containing 150 mM NMDG-MS, 5 mM MgCl₂, 10 mM EGTA, 10 mM citrate/Tris pH 4.7. The dose-dependence of 4-AP inhibition was examined by using the following protocol: from a holding potential of 0 mV, the voltage was stepped to voltages between −100 and +100 mV for 120 mS, in 20-mV increments, in the presence of varying concentrations of 4-AP at pH 4.2 bath solutions. For both 4-AP inhibition of Cs⁺ current H⁺ current recordings, the concentration of 4-AP in the bath solution was exchanged by rapid microperfusion (ALA Scientific Instruments). The voltage dependence of 4-AP inhibition was examined using the following protocol: from a holding potential of 0 mV, the voltage was stepped to voltages between −100 and +100 mV for 120 ms, in 20-mV increments. The voltage dependence of 4-AP dissociation was examined by holding at 0 mV, stepping to either −100 mV or +100 mV. After ∼3 s, 1 mM 4-AP was added to the bath solution by microperfusion for 3∼5 s and then removed. After a total of 20 s, the voltage was stepped to +100 mV or −100 mV and after ∼3 s, 1 mM 4-AP was added to the bath solution by microperfusion for 3∼5 s and then removed. The dissociation rate of 4-AP was calculated by fitting plateau followed by one-phase decay or one-phase association curve to data using GraphPad Prism (GraphPad Software). DMSO concentration was maintained at a constant concentration of 0.33% throughout the entire experiment.

The currents were recorded using Axon Digidata 1550B digitizer and Clampex 10.6 (Molecular Devices) and analyzed using AxoGraph X 1.7.6 (AxoGraph Scientific) and GraphPad Prism (GraphPad Software). Each experiment was performed in a unique cell.

Protein Expression and Purification.

TMEM175 was purified as described previously (9). Briefly, 1 mg of a plasmid encoding human TMEM175 with a C-terminal EGFP-tag fused via a short linker containing a PreScission protease site was mixed with 3 mg of PEI 25k (Polysciences) for 30 min and then used to transfect 1 L of HEK293S GnTi^– cells. After 24-h incubation at 37 °C, sodium butyrate was added to a final concentration of 10 mM, and cells were allowed to grow at 37 °C for an additional 48 to 72 h before harvesting. Cell pellets were washed in phosphate-buffered saline solution and flash frozen in liquid nitrogen. Expressed protein was solubilized in 2% lauryl maltose neopentyl glycol (LMNG, Anatrace), 20 mM Hepes pH 7.4, 150 mM KCl supplemented with protease-inhibitor mixture (1 mM PMSF, 2.5 μg/mL aprotinin, 2.5 μg/mL leupeptin, 1 μg/mL pepstatin A), and DNase. Solubilized protein was separated by centrifugation at 74,766 × g for 40 min, followed by binding to anti-GFP nanobody resin for 2 h. Anti-GFP nanobody affinity chromatography was performed by 20 column volumes of washing with buffer containing 0.1% LMNG, 20 mM Hepes pH 7.4, 150 mM KCl, 2 mM dithiothreitol (DTT), followed by overnight PreScission digestion, and elution with wash buffer. Eluted protein was further purified by size-exclusion chromatography on a Superdex 200 Increase 10/300 GL (GE Healthcare) in SEC buffer, 0.1% LMNG, 50 mM Tris pH 8.0, 150 mM KCl, 2 mM DTT, or 0.1% LMNG, 10 mM Hepes pH 7.4, 150 mM KCl, 2 mM DTT, for cryo-EM sample preparation and proteoliposome reconstitution, respectively. Approximately 0.5 mg of purified protein was obtained from each liter of cell culture. Peak fractions were pooled and concentrated to ∼4 mg/mL using CORNING SPIN-X concentrators (100-kDa cutoff).

Proteoliposome Reconstitution and Flux Assay.

POPE and POPG in chloroform (Avanti) were mixed in a ratio 3:1 (mg:mg) and dried under argon gas. The dried lipid mixture was solubilized in pentane and dried again under argon gas to remove residual chloroform. Dried lipids were then desiccated for 2 h under vacuum. Lipids were resuspended in 10 mM Hepes pH 7.4, 300 mM KCl to a final concentration of 10 mg/mL Unilamellar vesicles were formed by sonication and then solubilized using 8% (wt/vol) octyl maltoside. Full-length hTMEM175 purified in LMNG at a concentration of 1 mg/mL was mixed with the octyl maltoside-solubilized lipids and dialyzed using 25-kDa MWC bags (SpectraPor) in 10 mM Hepes pH 7.4, 300 mM KCl, 2 mM DTT for 5 d with daily exchange of dialysis buffer. After dialysis, harvested proteoliposomes were snap frozen in liquid nitrogen and stored at −80 °C until use. Proteoliposomes were rapidly thawed at 37 °C, sonicated for 5 s, incubated at room temperature for 2 to 4 h before use, and then diluted 100-fold into a flux assay buffer composed of 10 mM Hepes pH 7.4, 300 mM NaCl, 0.2 µM ACMA.

Data were collected on a SpectraMax M5 fluorometer (Molecular Devices) using Softmax Pro-6 software. ACMA excitation/emission wavelengths were 410/490 nm, respectively. Fluorescence intensity measurements were collected every 30 s. The ionophore CCCP (1 µM) and valinomycin (20 nM) were added at 180 s and at the end of experiments, respectively. The baseline measured by experiments with empty liposomes were subtracted. The rate was fit by plateau followed by one phase decay using GraphPad Prism (GraphPad Software). Collected data were analyzed using GraphPad Prism (GraphPad Software).

EM Sample Preparation and Data Acquisition.

Purified TMEM175 at a concentration of 4 mg/mL was mixed 1:10 with 10 mM 4-AP solubilized in SEC buffer with 1% DMSO. 4 to 5 μL was applied to glow-discharged Au 400 mesh QUANTIFOIL R1.2/1.3 holey carbon grids (Quantifoil), and then plunged into liquid nitrogen-cooled liquid ethane with an FEI Vitrobot Mark IV (FEI Thermo Fisher). Grids were transferred to a 300 keV FEI Titan Krios microscopy equipped with a K2 summit direct electron detector (Gatan). Images were recorded with SerialEM (26) in superresolution mode at 22,500×, corresponding to pixel size of 0.544 Å. Dose rate was eight electrons per pixel per second, and defocus range was 1.2 to 2.5 µm. Images were recorded for 8 s with 0.2-s subframes (total 40 subframes), corresponding to a total dose of 61 electrons/Ų.

Analysis of EM Images.

For analysis of EM images, 2,466 images were gain-corrected, Fourier-cropped by two, and aligned using whole-frame and local motion-correction algorithms by Motioncor2 (27) (1.088 Å per pixel). Whole-frame CTF parameters were determined using CTFfind 4.1.14 (28). Particles were automatically selected in Relion 3.0 using templates previously generated from 2D classification, resulting in 1,128,690 particles (29). False-positive selections and contaminants were excluded from the data using multiple rounds of heterogeneous classification in cryoSPARC v3.2 using the open and closed states, as well as several decoy classes generated from noise particles via ab initio reconstruction in cryoSPARC v3.2, resulting in a stack of 158,819 particles (30). After Bayesian polishing in Relion and local CTF estimation and higher-order aberration correction in cryoSPARC v3.2, a consensus reconstruction was determined at resolution of 2.9 Å (31). A second round of Bayesian polishing in Relion 3 using a pixel size of 0.85 Å and box size of 384 yielded an improved consensus reconstruction at 2.7 Å. The final reconstruction was subjected to density modification using the two unfiltered half-maps with a soft mask in Phenix (32).

Model Building and Coordinate Refinement.

The structures of open (PDB ID code 6WC9) human TMEM175 were docked into the density map in COOT and manually adjusted to fit the density (33). Densities corresponding to TM5 and TM6 (residues 174 to 251) were too poorly ordered and omitted from the model. The final models are composed of residues 30 to 173 and 254 to 476. Atomic coordinates were refined against the density modified map using phenix.real_space_refinement with geometric and Ramachandran restraints maintained throughout (34).

MD Simulations.

The simulations were based on high-resolution cryo-EM structures of the human TMEM175 channel in an uninhibited open state (PDB ID code 6WC9) and in complex with 4-AP (PDB ID code 8DHM). All simulations were conducted with NAMD 2.12 using the CHARMM36 force field for protein and lipids (35–38). Force field parameters for 4-AP compatible with CHARMM were derived using CGenFF (39) and are publicly available at https://github.com/Faraldo-Gomez-Lab-at-NIH/Download. All simulations were carried out at constant temperature (298 K) and semi-isotropic pressure (1 atm), using periodic boundary conditions and an integration time step of 2 fs. Long-range electrostatic interactions were calculated using PME, with a real-space cutoff of 12 Å. Van der Waals interactions were computed with a Lennard–Jones potential, cutoff at 12 Å, with a smooth switching function taking effect at 10 Å.

The specific protein construct studied includes residues 30 to 165 and residues 254 to 476. All ionizable sidechains were set in their default protonation state at pH 7, except for H57, which was protonated on account of its proximity to D279, E282, and D283. K⁺ ions and water molecules originally modeled within the pore in the cryo-EM structure were removed. The protein constructs (and bound inhibitor) were embedded in a pre-equilibrated hydrated palmitoyl-oleoyl-phosphatidyl-choline (POPC) lipid bilayer using GRIFFIN, and enclosed in a periodic orthorhombic box of ∼100 × 100 × 110 Å in size (40). The resulting simulation system for the open channel contains 222 POPC lipids, 24,342 water molecules, 49 Cl⁻ ions, and 43 K⁺ ions; that for the complex with 4-AP contains 222 POPC lipids, 24,336 water molecules, 50 Cl⁻ ions, and 43 K⁺ (that is, a salt concentration of 100 mM plus counterions to neutralize the net charge of the protein complex). The simulation systems were equilibrated following a staged protocol comprising a series of MD trajectories wherein positional and conformational restraints acting on the protein structure are gradually weakened over 100 ns.

Following completion of this protocol, two independent trajectories of 1 μs each were calculated for the inhibited state, free of any configurational restraints, and initiated in different configurations extracted from the final stage of the equilibration; a trajectory of 1 μs was also calculated for the apo state, analogously. In both cases the TM voltage was set to zero. Four additional trajectories of 250 ns each were generated for the 4-AP bound state, with a voltage V_z = −500 mV applied across the membrane (positive on the luminal side). Each trajectory was initiated with a different configuration, following a 100-ns equilibration at −300 mV. To implement this TM potential, an electric field was applied perpendicularly to the membrane plane, directed toward the cytosolic side, of magnitude E_z = V_z/L_z = −0.1043 kcal/mol Å⁻¹ e⁻¹, where L_z = 110 Å is the average length of the box in that direction (1 k_BT = 0.592 kcal/mol = 25.8 mV × e).

Descriptor definitions for data analysis are indicated throughout the manuscript. In Fig. 4, the so-called “contact number” between two groups of atoms in each simulation snapshot ${\mathbf{X}}_{\text{k}}$ is defined as:

$$\text{N} \left({\mathbf{X}}_{\text{k}}\right) ={\displaystyle {\sum }_{\text{i}}{\displaystyle {\sum }_{\text{j}}\frac{1-{\left({\text{d}}_{\text{ij}}\left({\mathbf{X}}_{\text{k}}\right)/{\text{d}}_0\right)}^{100}}{1-{\left({\text{d}}_{\text{ij}}\left({\mathbf{X}}_{\text{k}}\right)/{\text{d}}_0\right)}^{200}}}}$$

where the i and j indexes denote a pair of atoms in each group, and ${\text{d}}_{\text{ij}}\left({\mathbf{X}}_{\text{k}}\right)$ denotes the distance between them in snapshot ${\mathbf{X}}_{\text{k}}$. For the evaluation of contacts between 4-AP and water oxygen atoms, ${\text{d}}_0$ was set to 3.5 Å. In the evaluation of contacts with the hydroxyl oxygen of Ser45, ${\text{d}}_0$ was set to 4.5 Å. In both cases ${\text{d}}_0$ is the minimum distance that approximately encompasses the first peak in the radial-distribution-function characteristic of each type of contact. To quantify the shape of the transmembrane potential that results from the external field E_z, when 4-AP is bound to TMEM175 (Fig. 6B), we first calculated time-averages of the electrostatic potential across the simulation system Φ (X), both for the trajectory obtained at 0 mV and for that at 500 mV. These potentials were calculated with the PMEpot plug-in in VMD (41), using an Ewald factor of 0.25 Å⁻¹, and mapped onto a lattice of 200 × 200 × 224 grid-points. From these 3D maps, we then derived one-dimensional profiles along the membrane perpendicular, Φ₀ (z) and Φ₅₀₀ (z), by averaging all potential values over xy planes. The transmembrane potential Φ_tm (z) induced by the external field was then derived as Φ_tm (z) = − E_z × z + [Φ₅₀₀ (z) − Φ₀ (z)]. In this expression, the second term represents the linear trend that would be expected for a homogeneous medium, while the first term quantifies the response of the actual, nonhomogeneous molecular system to the applied voltage, relative to the zero-voltage condition.

Data, Materials, and Software Availability

Cryo-EM maps and atomic coordinates for 4-AP bound TMEM175 have been deposited with the European Molecular Data Bank (EMDB, https://www.ebi.ac.uk/emdb/) and Protein Data Bank (PDB, https://www.rcsb.org/) (accession nos. EMDB-27436 (42) and PDB 8DHM (43)) Datasets from the PDB used in this study include 2A79 (44), 6WC9 (45), 7PHL (46), 7SSV (47), and 7UNL (48). Force field parameters for 4-AP compatible with CHARMM are publicly available at https://github.com/Faraldo-Gomez-Lab-at-NIH/Download. Source data are provided with the paper. All other data are included in the manuscript and/or SI Appendix.