Allosteric coupling between PIP₂ and Ca²⁺ binding sites gates TMEM16A channels

Significance

TMEM16A is a Ca²⁺ -activated chloride channel expressed in various tissues, where it regulates electrical excitability, cell volume, ion homeostasis, and solute transport. Dysfunction of TMEM16A is linked to conditions including hypertension, asthma, and cancer, making it a potential drug target for a range of disorders. Besides elevated intracellular Ca²⁺ levels, TMEM16A activation requires interactions with phosphatidylinositol 4,5-bisphosphate (PIP₂). However, the role of PIP₂ in TMEM16A gating and Cl⁻ permeation remains unclear. Here, we show that both the phosphate headgroups and acyl chains of PIP₂ interact with the α4 helix of TMEM16A, opening an electrostatic gating ring and stabilizing the Cl⁻ conduction pathway. These findings clarify PIP₂’s role in TMEM16A activation and provide insights for targeted drug development.

Abstract

TMEM16A channels conduct Ca²⁺-activated Cl⁻ currents that underlie essential physiological processes including epithelial secretion, smooth muscle contraction, and sensory transduction. Channel activation requires both intracellular Ca²⁺ and the signaling phospholipid phosphatidylinositol 4,5-bisphosphate (PIP₂), yet the molecular basis of this dual regulation has remained unclear. Using gating molecular-dynamics simulations and structure-guided electrophysiology, we show that PIP₂ and Ca²⁺ cooperatively gate TMEM16A through an allosterically coupled electrostatic network centered on the α4 helix. Specific PIP₂ headgroup phosphate interactions are essential for coupling Ca²⁺ binding to channel opening, while the PIP₂ acyl chains engage hydrophobic surfaces of the helix to stabilize the open conformation. Disrupting either component of this lipid–protein interface reduces apparent PIP₂ affinity and impairs activation, whereas long-chain PIP₂ fully restores wild-type activity. These interactions act in concert with Ca²⁺-dependent structural rearrangements that widen the conduction pathway and enable Cl⁻ permeation. Our findings establish that both the headgroup phosphates and acyl chains of PIP₂ play indispensable and complementary roles in TMEM16A gating. This mechanism defines a cooperative lipid–ion activation process that provides a general framework for understanding phosphoinositide regulation of ion channels and offers opportunities for structure-based design of TMEM16A modulators.

TMEM16A, also known as ANO1, is the prototypical member of the anoctamin family of transmembrane proteins and functions as a Ca²⁺-activated chloride channel (1–3). In humans, TMEM16A regulates the electrical excitability of muscle and multiple types of neurons. The channel is also expressed in epithelial tissues where it plays a key role in regulating cell volume, ion homeostasis, and solute transport (4). Increasingly, quantitative expression profiling in pathophysiological studies is establishing a link between changes in anoctamin activity, and a variety of pathophysiological processes, highlighting TMEM16A channels as promising druggable targets for several diseases, from hypertension, moyamoya disease to various types of cancer (5, 6).

TMEM16A channels form as dimers of subunits that each contain 10 transmembrane helices (α1-10) and a putative Cl⁻ permeation pathway. Under physiological conditions, the channels activate when the intracellular Ca²⁺ concentration exceeds approximately 150 nM, with Ca²⁺ binding at two well-characterized sites (4, 7–9). Despite these foundational insights from cryo-electron microscopy (cryo-EM) structures, the mechanistic basis for channel gating and the permeation pathway are not resolved.

Several functional studies have demonstrated that, in common with many types of membrane proteins, and most types of ion channels, TMEM16A gating is dependent on the presence of PIP₂ in the plasma membrane (10–12). Although PIP₂ constitutes only a minor fraction of the plasma membrane phospholipid pool, it is increasingly recognized as a master regulator of ion channel function. Despite this pivotal role in membrane physiology, the precise mechanisms by which PIP₂ regulates channel gating, including that of TMEM16A, are often unclear (13–15). In silico docking and mutagenesis studies have identified several potential PIP₂-binding pockets on TMEM16A; however, the functional significance of these sites is ambiguous (16). Furthermore, it is unclear whether these binding sites serve orthosteric roles, directly mediating channel activation, or act allosterically to modulate gating indirectly. This knowledge gap limits our understanding of the channel’s activation mechanics and poses a significant challenge to the rational design of drugs targeting TMEM16A, hindering therapeutic advancements aimed at modulating its activity. Here, we identify the PIP₂-binding interactions that directly govern how TMEM16A channels are activated. We characterize how PIP₂ interacts with the α4 and α6 helices of TMEM16A through both its phosphate headgroups and acyl chains revealing the mechanism that underpins channel gating. When Ca²⁺ binds to TMEM16A, PIP₂ reorients in its binding pocket to facilitate the opening of an electrostatic gating ring within the channel core and stabilize the opening of the extracellular mouth of the ion conduction pathway.

Results

Identification of the PIP₂ Interface on TMEM16A.

Using induced fit docking, we identified four distinct PIP₂-binding pockets (sites-1 to 4) in the model of mouse TMEM16A constructed from the published structure (PDB: 5OYB) (Fig. 1-site 1 and SI Appendix, Fig. S1-sites 2 to 4). Sites 2 to 4 overlap with PIP₂ interaction interfaces that have been proposed by others (10, 11, 16, 17). Next, we studied each system using gating molecular dynamics (GMD) simulations, a computation approach to study ion channel gating dynamics and ion permeation (18, 19). Using GMD, we observed Cl⁻ ions permeation through the channel only when PIP₂ interacted with the site-1 binding pocket. Thus, during 2,000 ns GMD simulations, the site-1 system passed 100 ± 48 Cl⁻ per µs (mean ± SD, n = 3) when Ca²⁺ and PIP₂ were both bound to the channel. No permeation events were observed in the apo-channel or the PIP₂ alone systems, while relatively fewer permeation events were seen with Ca²⁺ alone (Fig. 1B). The Cl⁻ ions permeated the channel through a dedicated pathway formed by a series of basic residues including the α4 (R535, R562, and K566) and α6 (K645 and K661) helices of each subunit (Fig. 1A). Consistent with the requirement for both Ca²⁺ and PIP₂ to gate the channel, principal component analysis (PCA) revealed large conformational changes in the extracellular domain of the α4 helix when both regulators were present. The pronounced increase in the Root Mean Square Fluctuation (RMSF) in the motion of helix α4 contrasts with the relatively limited motion of helices α1-3 and α5-7, both in the presence and absence of Ca²⁺ and PIP₂ (Fig. 1 C–E). A marked decrease in the RMSF of the α6 helix was also observed in the presence of Ca²⁺ (Fig. 1E_ii), which indicated decreased flexibility of the α6 helix.

Fig. 1.

A Ca²⁺-coupled allosteric activation site on TMEM16A for PIP₂. Ca²⁺-bound TMEM16A channels (PDB: 5OYB) activate and pass Cl⁻ in silico when PIP₂ is docked to site-1. (A) Left: The Cl⁻ permeation pathway through a TMEM16A subunit; PIP₂ and Ca²⁺ are removed for clarity. Right: Example raster plots showing Cl⁻ permeation events. (B) Mean ± SD for three simulation replicates measuring Cl⁻ flux per µs in different TMEM16A systems (apo, +PIP₂, +Ca²⁺, and +PIP₂+Ca²⁺). When present, PIP₂ is in site-1. (C) PCA showing increased motion of the external α4 helix in the PIP₂ + Ca²⁺ system and decreased motion of the internal α6 helix in Ca²⁺-containing systems. (D) Summary of RMSFs across the length of helices α1-7. (E_i) The fluctuation in α4 motion requires Ca²⁺ and PIP₂. (E_ii) Movement of α6 is subdued in the Ca²⁺-containing systems. (F_i) Binding pose of PIP₂ in site-1 in the absence and, (F_ii) presence of Ca²⁺. A more complete version of this model is shown in SI Appendix, Fig. S2. (G) Distance between the P5–phosphate of PIP₂ and R665. (H) Distance between the P1-phosphate of PIP₂ and K661 and K668. (I) Distance between the P4-phosphate of PIP₂ and R562 and, (J) K566.

Under basal physiological conditions, PIP₂ is available in the plasma membrane to interact with TMEM16A. However, the channel exhibits only low activity until Ca²⁺ ions bind. To investigate the role of PIP₂ in the gating process, we analyzed the trajectories of interactions between the channel and PIP₂ in the presence and absence of bound Ca²⁺. In the absence of Ca²⁺, PIP₂ forms stable electrostatic interactions within 5 Å of basic residues on α6. Specifically, the phosphate at position 5 (P5) of the PIP₂ inositol ring interacts with R665, while the phosphate at position 1 (P1) interacts with K661 (Fig. 1 F_i–H and SI Appendix, Fig. S2). Upon Ca²⁺ binding, the α4 helix reorients and PIP₂ adopts a new binding pose. The resulting sequence of interactions is evident in the GMD trajectories. There is an immediate disengagement of P5 and R665, followed by a more gradual loss of contact between P1 and K661, which is fully disrupted approximately 700 ns into the simulation (Fig. 1 G and H). At the same time that the P1–K661 interaction breaks, the K668 residue comes within 10 Å of P1 (Fig. 1H). In place of these interactions, the P4-phosphate immediately engages R562 on α4 and subsequently K566 after ~800 ns (Fig. 1 F_ii–J and SI Appendix, Fig. S2). The Ca²⁺-dependent conformational transitions were consistently observed across three independent 2,000 ns simulations (SI Appendix, Fig. S3).

To validate this dynamic model, we performed patch-clamp recordings from TMEM16A channels with point mutations in the putative PIP₂ interaction interface. Mouse TMEM16A (GenBank NM_178642.6) in the pEGFP-N1 expression plasmid was transiently transfected into HEK293T cells. Mean whole-cell currents were 174 ± 22 pA/pF (mean ± SD, n = 80) at +80 mV with 455 nM free Ca²⁺ in the intracellular recording buffer (Fig. 2A) (12). We designed a number of substitutions to test both the electrostatic and structural contributions of basic residues predicted to coordinate the PIP₂ headgroup. Alanine substitutions (R562A and K566A) remove both charge and hydrogen-bonding capacity and were expected to disrupt PIP₂ coordination and reduce current. Glutamine substitutions (R562Q and K566Q) neutralize the charge while preserving side-chain length and polar character, allowing assessment of whether hydrogen-bonding or steric complementarity can compensate for the lost positive charge.

Fig. 2.

Functional consequences of PIP₂ interactions at the Ca²⁺-coupled allosteric activation site. mTMEM16A was expressed in HEK293T cells and currents were recorded in whole-cell or excised inside-out patch mode. The electrophysiological data are presented as mean ± SD and represent at least eight cells per group obtained from three independent biological replicates of each experiment. Statistical differences between the groups was determined by two-tailed, unpaired, Student’s t test. *** denotes P < 0.001; NS is not significant. (A) Left: Example whole-cell TMEM16A currents evoked by a 500 ms step to 80 mV. Right: Summary mean current densities from TMEM16A variants. (B) Left: The time course of whole-cell TMEM16A current magnitude following photoactivation (blue box) of the light activated phosphatases CRY2-5_OCRL or CRY2-PJ. Right: Summary plots of mean current remaining following activation of (Upper) CRY2-5_OCRL or (Lower) CRY2-PJ. (C) Summary of TMEM16A channel current remaining following photoactivation of CRY2-PJ with (i) diC8-PIP₂; (ii) diC8-PI(4)P, or (iii) diC8-PI(5)P in the recording pipette. (D) Excised inside-out patch experiments to study the apparent-affinity of TMEM16A for diC8-PIP₂ (i) Example TMEM16A currents in an excised inside-out patch showing depletion by poly-L-lysine (PLL, 100 µg/mL) and concentration-dependent recovery with diC8-PIP₂. The example shows application 10 nM 1), 1 µM 2), and 100 µM 3) diC8-PIP₂. The red dashed line indicates 0 pA, the black dashed line indicates the original current magnitude. (ii) Example K566Q current recorded in an excised patch experiment showing concentration-dependent recovery of current following PLL with application of 10 nM 1), 1 µM 2), and 100 µM 3) diC8-PIP_2. The scale bar applies to (i and ii). (iii) diC8-PIP₂ recovered channel activity in a concentration-dependent manner, with an apparent-affinity (EC₅₀) of 2.3 µM for the WT and 0.3 µM for K566Q.

K566 substitutions enhanced channel activity, suggesting that partial relaxation of the electrostatic network at this site favors opening. K566A and K566Q channels displayed 2.5-fold higher mean currents compared with the wild type (WT) (Fig. 2A). R562Q channels, in contrast, exhibited WT-like currents, indicating that a polar side chain can maintain gating, whereas R562A channels were nonconductive, showing that complete loss of polarity at this position disrupts activation.

To investigate the basis for the enhanced currents observed with K566 substitutions, we next examined whether these effects could be attributed to differences in surface expression. We used total internal reflection fluorescence (TIRF) microscopy to quantify the membrane-localized fluorescence of eGFP-tagged TMEM16A variants in HEK293T cells. The mean TIRF intensities of R562A, R562Q, K566A, and K566Q channels were not significantly different from those of WT TMEM16A (SI Appendix, Fig. S4). Likewise, all additional mutants tested in subsequent functional experiments exhibited comparable surface expression levels to the WT (SI Appendix, Fig. S4), indicating that their altered macroscopic currents are not due to changes in surface expression.

To determine whether the altered macroscopic currents in site-1 mutants reflect changes in ion permeation or in gating, we next analyzed current–voltage (I–V) relationships for WT TMEM16A and the K566A, K566Q, R665Q, and K668Q variants. All constructs exhibited rectifying I–V curves like WT, with reversal potentials close to the predicted E_Cl (~−4 mV) under our recording conditions (SI Appendix, Fig. S5). The principal effect of these substitutions was to change the current amplitude at depolarized voltages, without detectable shifts in reversal potential or overall I–V shape. These data indicate that the site-1 mutations do not produce major changes in Cl⁻ selectivity and instead primarily modulate channel activation.

To further probe the mechanism underlying the enhanced currents observed with K566 substitutions, we investigated how TMEM16A interacts with specific phosphates of the PIP₂ headgroup using a light-activated phosphoinositide (PI) phosphatase system. This optogenetic platform is based on photoactivation (~440 to 490 nm) of Cryptochrome 2 (CRY2) and its binding partner, CIBN (20). By fusing CIBN to the membrane anchor CAAX, photostimulation recruits CRY2 and its fused cargo to the inner leaflet of the plasma membrane (21, 22). Prior studies using PIP₂-sensitive ion channels or fluorescent biosensors (e.g., the PH domain of phospholipase Cδ) have shown that CRY2–phosphatase constructs translocate to the membrane within seconds, rapidly dephosphorylating PIP₂. Although PIP₂ can be resynthesized within minutes, continuous illumination maintains depletion (23, 25, 26).

We first examined the effects of photoactivation of CRY2 fused to pseudojanin (CRY2–PJ), a tandem phosphatase combining the 5-phosphatase INPP5E and the 4-phosphatase Sac1, which depletes PIP₂ by sequential conversion to PI(4)P and then PI (27). Blue-light activation of CRY2–PJ reduced whole-cell currents by ~95% within three min in cells expressing WT TMEM16A (Fig. 2B). In contrast, activation of the P5-specific phosphatase CRY2–5_OCRL, which converts PIP₂ to PI(4)P, did not reduce WT TMEM16A currents. These findings indicate that at least one phosphate at the 4′ position is required to maintain channel gating, consistent with our in silico model of PIP₂-dependent activation. CRY2-PJ also reduced whole-cell currents from K566A and K566Q channels by ~95%. However, in contrast to the WT, CRY2-5_OCRL also reduced currents from K566A and K566Q channels by ~95%, suggesting that activity of these TMEM16A variants is not supported by PI(4)P (Fig. 1B).

Following the observation that photoactivation of CRY2–PJ depleted TMEM16A currents, we next sought to identify which phosphate groups on the PIP₂ headgroup are essential for channel activation. To this end, we photoactivated CRY2–PJ while including 50 µM of soluble, short-chain (diC8) PIP₂, PI(4)P, or PI(5)P in the recording pipette. In WT TMEM16A channels, diC8–PIP₂ preserved 55 ± 5% (mean ± SD, n = 10) of the current, diC8–PI(4)P preserved 75 ± 6%, and diC8–PI(5)P failed to preserve any current, consistent with our model predicting that 4′-phosphate interactions are the principal drivers of activation (Fig. 2C).

For TMEM16A-K566A and K566Q channels, diC8–PI(4)P produced no rescue, while diC8–PI(5)P preserved ~60% of activity and diC8–PIP2 rescued ~50% of the current (Fig. 2C). These results suggest that in the K566 variants, P5-mediated interactions dominate channel activation. Supporting this conclusion, combined photoactivation of CRY2–5_OCRL and CRY2–4Sac1 confirmed sensitivity to dephosphorylation at the 5′ position (SI Appendix, Fig. S6A). Furthermore, mutations in α6-helix residues predicted to coordinate the P1 and P5 phosphates (R665Q and K668Q, respectively) were rescued by PI(5)P or PIP₂ but not by PI(4)P, again implicating 5′-phosphate interactions in activation (SI Appendix, Fig. S6 C–E).

Together with the TIRF and I-V analyses, these optogenetic phosphatase and lipid-rescue experiments indicate that the larger macroscopic currents in K566A and K566Q primarily reflect altered PIP₂-dependent gating, rather than changes in surface expression or gross ion selectivity.

To gain structural insight into how the K566Q substitution favors P5-mediated interactions and enhances channel activation, we next analyzed PIP₂–channel contacts in a molecular model of TMEM16A-K566Q in the presence of PIP₂ and bound Ca²⁺ (SI Appendix, Fig. S6). We quantified time-averaged contact probabilities between each PIP₂ phosphate (P1, P4, P5) and mostly basic residues on α4 (R562, K/Q566) and α6 (K661, R665, K668). In the WT channel, P4 interacts with both R562 and K566, whereas P5 and P1 make less frequent contact with R665 and K668. Substitution of K566 with glutamine disrupts the electrostatic interaction with P4 and is predicted to reorient the PIP₂ headgroup within its binding pocket. In the K566Q simulations, contacts between P4 and R562 and between P5 and R665 increase, and P1 engages K668 more persistently, generating a more extensive electrostatic bridge between α4 and α6 (Fig. 1 G–I and SI Appendix, Fig. S6 F–H). Although we cannot resolve these individual side-chain contacts experimentally, this reconfigured contact pattern provides a structural explanation for the enhanced PIP₂ sensitivity of K566Q observed in our functional assays. Similar enhancements in current density were obtained with R665A/Q and K668A/Q (SI Appendix, Fig. S6H), consistent with an important role for the α4–α6 contact network in site-1 gating.

Finally, to test whether this enhanced activity reflected a higher apparent affinity for PIP₂ we recorded currents from off-cell inside-out patches. Endogenous PIP₂ was first removed by poly-L-lysine, abolishing current, followed by addition of exogenous diC8-PIP₂ to the cytosolic face. WT TMEM16A channels were activated with an EC₅₀ of 2.3 µM, consistent with prior reports (17). In contrast, TMEM16A-K566Q channels were activated with an EC₅₀ of 0.3 µM, indicating an ~eightfold increase in apparent affinity for PIP₂ (Fig. 2D).

TMEM16A-R562A channels showed no rescue by any diC8 lipid, underscoring the importance of a polar side chain at this position (Fig. 2C). In contrast, R562Q channels retained partial regulation: diC8–PIP₂ preserved 54 ± 9% of current, diC8–PI(4)P 44 ± 3%, and diC8–PI(5)P 82 ± 5%, indicating that 5′-phosphate interactions become predominant when the positive charge at R562 is neutralized.

Together, the electrophysiological and modeling data indicate that the enhanced currents in TMEM16A-K566Q channels arise from a reoriented PIP₂-binding configuration that stabilizes multiple simultaneous interactions with the α4 and α6 helices, thereby increasing the apparent affinity for PIP₂ and promoting channel opening. Based on this analysis of the K566 variants, we anticipate that R562A fails to coordinate PIP₂ in a configuration that supports activation, consistent with its complete loss of current.

PIP₂ Acyl Chains Interact with α4.

In addition to analyzing interactions between the phosphatidylinositol headgroup of PIP₂ and TMEM16A, we examined how the two acyl chains of PIP₂ engage residues along the α4 helix in both the Ca²⁺-free and Ca²⁺-bound systems of the site-1 GMD model. To generate these complexes, a full PIP₂ molecule was first positioned in the canonical binding pocket identified in Fig. 1, where the headgroup coordinates the basic residues. The complete lipid, including both acyl tails, was then subjected to molecular-dynamics equilibration within an explicit lipid bilayer. After a brief restrained minimization to remove steric clashes, the restraints were released, allowing the acyl chains to explore freely the surrounding hydrophobic cavity around α4. In these simulations, the acyl tails repeatedly contacted residues N546–I551, a segment that overlaps the previously described hydrophobic gate defined by I550/I551 (28, 29), suggesting that PIP₂ tail interactions at this gate could modulate open probability. Thus, while the headgroup remained anchored in the described binding pocket, the tails were unconstrained and able to adopt conformations dictated by local lipid–protein interactions.

During the simulations, carbons 6 to 12 of the arachidonyl (sn-2) tail repeatedly contacted residues N546–I551 of α4, with I,550 showing persistent van der Waals interactions near carbons 9 to 11 (Fig. 3 A and B and SI Appendix, Fig. S7). These contacts were stronger and more stable in the Ca²⁺-bound trajectories, suggesting that calcium-dependent conformational tightening enhances PIP₂–channel coupling. Consistent with this, distance-distribution histograms for individual residues (N546, L547, V548, V549, I550, and I551) show sharper peaks at shorter distances (2 to 6 Å) in the presence of Ca²⁺ and broader, lower-amplitude distributions in its absence (SI Appendix, Fig. S7A), indicating closer and more persistent acyl-chain engagement under Ca²⁺-bound conditions.

Fig. 3.

PIP₂ acyl-chain contacts with α4 support allosteric gating. Site-1 GMD studies show the PIP₂ acyl chain carbon interactions with α4 residues N546 through to I551. The electrophysiological data are presented as mean ± SD for at least eight independent recordings per condition obtained from three independent biological replicates of each experiment. Statistical differences between the groups were determined by two-tailed, unpaired, Student’s t test. * denotes P < 0.05, **P < 0.01 and ***P < 0.001; NS is not significant. (A) Snapshot from the site-1 GMD simulation showing examples of the shortest interaction distances between contact residues and the PIP₂ carbon observed during the trajectory. (B) Trajectory plots show that interactions are stabilized in the presence of PIP₂ and Ca²⁺. (C) Summary of current magnitudes when residues in the α4 are substituted for alanine. (D) Summary of current rescue in excised inside-out patches by the application of (i) diC8-PIP₂ or (ii) long-chain PIP₂ according to the experimental regime described in Fig. 2D; (iii) The apparent affinity of TMEM16A-I550A to diC8-PIP₂ is 11 µM, compared to 2.3 µM for the WT channel.

The functional consequences of these interactions were examined by whole-cell patch-clamp recordings from alanine-substitution mutants spanning A542–L553 (Fig. 3C). N546A and L547A exhibited no measurable whole-cell current, and in excised inside-out patches neither diC8-PIP₂ nor long-chain (LC) PIP₂ rescued channel activity, consistent with a complete loss of productive PIP₂ engagement. The V548A, V549A, and I550A substitutions each reduced whole-cell current amplitude to less than half that of the WT, whereas I551A retained approximately 70% of the WT current. In excised patches, both diC8- and LC-PIP₂ restored currents to roughly half the WT level, but only I550A displayed clear chain-length dependence: diC8-PIP₂ produced partial recovery, whereas LC-PIP₂ fully restored WT current levels, indicating that long-chain hydrophobic contacts at I550 are essential for full stabilization. Concentration–response analysis further revealed a right-shifted apparent affinity for I550A: The EC₅₀ for diC8-PIP₂ was 11 µM compared to 2.3 µM for the WT channel (Fig. 3D_iii), consistent with weakened binding and reduced coupling efficiency. By contrast, L552A and L553A (residues located closer to the PIP₂ headgroup) showed WT-like currents in both whole-cell and excised-patch assays and were rescued equally well by diC8- or LC-PIP₂. This is in line with the distance-distribution analysis, which showed that these residues maintain frequent but less chain-length-specific contacts with the proximal carbons of the acyl tails (SI Appendix, Fig. S7).

These results define the N546–I551 region of α4 as a critical determinant of acyl-chain-mediated gating. The pattern of rescue behavior parallels the predicted tail-contact map from simulation: Residues that engage the mid-tail carbons (e.g., I550) display chain-length-specific rescue, whereas residues nearer the headgroup (e.g., I552–L553) do not. This interpretation is consistent with previous cryo-EM and electrophysiological studies showing that I550 and I551 form part of the TMEM16A hydrophobic gate (28, 29). Our findings extend those results by demonstrating that the same residues also participate in lipid-dependent stabilization of the conductive conformation, linking the hydrophobic gate to the acyl-tail anchoring of PIP₂.

An Electrostatic Gating Ring in the Core of TMEM16A.

The precise location and molecular composition of the limiting constriction, or “gate,” within the TMEM16A permeation pathway remain incompletely defined. Previous cryo-EM and functional studies (30) identified several positively and negatively charged residues, most notably K588, K645, and E705, as forming part of an internal electrostatic environment that shapes the pore and contributes to ion conduction. However, those studies did not fully identify a “gate” and its relationship to PIP₂ and Ca²⁺.

In our GMD simulations of the Ca²⁺-bound site-1 system, we observed that D554 (α4), K588 (α5), K645 (α6), and E705 (α7) form a symmetric ring of charged residues located at the narrowest region of the conduction pathway (Fig. 4A). Part of this putative electrostatic gate is E705 that has been shown to coordinate Ca²⁺. D554 occupies a strategic position between two key PIP₂-interacting residues on α4, proximal to I551, which engages the acyl chains, and R562, which coordinates the P4 phosphate of the headgroup, suggesting that PIP₂ binding could directly influence the electrostatic architecture of this constriction.

Fig. 4.

An electrostatic gating ring mediates allosteric activation of TMEM16A. The electrophysiological data are presented as mean ± SD for at least eight independent recordings per condition obtained from three independent biological replicates of each experiment. Statistical differences between the groups were determined by two-tailed, unpaired, Student’s t test. ** denotes P < 0.01, ***P < 0.001, **** P < 0.0001. (A_i) Snapshot from the site-1 GMD study showing the relative position of D554 (α4), K588 (α5), K656 (α6), and E705 (α7) in the presence of PIP₂ and Ca²⁺. (A_ii) Snapshots of the gating ring showing the distances between the residues in the presence of PIP₂ (Left) and PIP₂ with Ca²⁺, where Cl⁻ is shown permeating the channel (Right). (B) Mean current-voltage plots for WT TMEM16A and D554 variants. (C) Mean current-voltage plots for WT TMEM16A and K588 variants. (D) Mean current-voltage plots for WT TMEM16A and K645 variants. (E) Mean current–voltage plots for WT TMEM16A and E705 variants.

In simulations containing PIP₂ but no Ca²⁺, the aperture formed by these residues remained too narrow for Cl⁻ permeation. Upon Ca²⁺ binding, E705 reoriented to coordinate the cation, which in turn rotated K645 outward, increasing the pore diameter and allowing anion passage. These transitions point to a cooperative electrostatic gating ring in which charged residues from four consecutive helices couple the movements of α4 (PIP₂ sensitive) and α7 (Ca²⁺ binding) to regulate ion access (Fig. 4A).

To test this hypothesis experimentally, we performed mutagenesis of each residue predicted to contribute to the gating ring (Fig. 4 B–E). Substitutions preserving the native charge (e.g., D554E, K588R, K645R, E705D) produced either no effect or modest current reductions (<30%). In contrast, reversing the charge of these residues nearly abolished current (>90% reduction). Neutralizing the side chains by substitution with alanine or glutamine reduced current magnitude by approximately half, consistent with partial disruption of the electrostatic coupling network. Importantly, these effects were not attributable to altered expression or trafficking, as TIRF microscopy confirmed comparable surface abundance for all variants (SI Appendix, Fig. S4).

Taken together, these findings are consistent with prior evidence that K588, K645, and E705 form an electrostatic constriction within the permeation pathway (30) and extend that model by demonstrating that D554, located on the PIP₂-responsive α4 helix, completes this network as a fourth coordinating element. We therefore propose that TMEM16A gating arises from an electrostatically coupled, four-helix ring in which PIP₂-induced motion of α4 modulates pore diameter through its linkage to the Ca²⁺-binding region. These results refine the existing model of TMEM16A pore activation by providing a structural explanation for how PIP₂ and Ca²⁺ jointly stabilize the conductive conformation through the same electrostatic network.

The Extracellular Mouth of the TMEM16A Pore.

Last, we investigated the effect of PIP₂ on the α4 extracellular domain of the channel by examining the trajectory of interactions between the flexible α4 and the α6 helices in the Ca²⁺-bound site-1 system. PCA revealed that the external portion of α4 undergoes significant fluctuations in motion upon Ca²⁺ binding (Fig. 1 C–E). In the absence of PIP₂, the α4-residue R535 can form a salt bridge with E633 on α6. In the presence of PIP₂ alone these two residues stably move away from each other precluding their interaction. Moreover, when both PIP₂ and Ca²⁺ are present, the α4 helix pivots away from α6 by approximately 20 Å (Fig. 5 A and B). To quantify the impact of this rearrangement on anion access, we measured the time-averaged number of Cl⁻ ions within 5 Å of basic residues lining the extracellular vestibule. In the Ca²⁺ + PIP₂ condition, this analysis revealed increased Cl⁻ occupancy around R535, N533, and N525 compared with Ca²⁺ alone or PIP₂ alone (Fig. 5C), suggesting that PIP₂ binding at site 1 promotes a more electrostatically favorable path for Cl⁻ approach to the permeation pathway.

Fig. 5.

PIP₂-dependent opening of the external TMEM16A pore. The electrophysiological data are presented as mean ± SD for 8 to 12 independent recordings per condition obtained from three independent biological replicates of each experiment. Statistical differences between the groups was determined by two-tailed, unpaired, Student’s t test. *** denotes P < 0.001. (A) Model showing a salt bridge between R535 (α4) and E633 (α6). (B) Trajectory plot showing disruption in the R535 to E633 interaction with PIP₂ and Ca²⁺ bound to the channel. (C) Summary of the number of Cl⁻ ions retained within 5 Å of the residue noted over 2,000 ns simulations. (D) Whole-cell patch clamp recording of TMEM16A channel variants. (i) Example whole-cell currents for WT TMEM16A and TMEM16A-R535A channels. (ii), Summary of mean current-densities at +80 mV for the WT and channel variants shown. (E) Summary of current rescue in off-cell patches by the application of phosphoinositide variants to WT TMEM16A and TMEM16A-R535 channels. (F) Schema showing the location of the major residues on the α4 helix that interact with PIP₂ or form part of the gating ring structure (i.e., D554) or the entry point of the permeation pathway (e.g., R535).

Supporting these modeling predictions, whole-cell patch-clamp recordings showed that mutations at residues where Cl⁻ occupancy is predicted to increase significantly enhanced macroscopic Cl⁻ currents through TMEM16A channels (Fig. 5D). We focused on R535A as a representative vestibule mutant and examined its PIP₂ dependence in excised, inside-out patches. As for WT channels, CRY2–pseudojanin activation and supplementation with diC8-PIP₂ reduced and rescued R535A currents respectively, indicating that the mutant remains PIP₂ dependent. However, R535A displayed larger currents at depolarized voltages than WT. To test whether this reflects altered permeation properties, we analyzed I–V relationships for WT and R535A channels. R535A exhibited larger outward currents but retained the rectification and reversal potential characteristics of the WT channels (SI Appendix, Fig. S5), indicating preserved Cl⁻ selectivity. TIRF imaging also revealed comparable surface abundance for R535A and WT channels (SI Appendix, Fig. S4), arguing against altered surface expression. Last, activation of the R535A mutant by PIP₂ in inside-out patches showed indistinguishable PIP₂ dependence, arguing against this mutant having a direct effect on PIP₂ sensitivity. We therefore interpret the increased currents in vestibule mutants as being consistent with an allosteric PIP₂-dependent facilitation of Cl⁻ entry into the pore. Because TMEM16A has a very small single-channel conductance in the low-picosiemens range, our whole-cell recordings cannot fully distinguish changes in open probability from subtle effects on unitary conductance or access resistance (28, 31).

Discussion

In this study, we identify and characterize the molecular interactions by which PIP₂ activates TMEM16A channels. By combining gating molecular-dynamics (GMD) simulations with targeted mutagenesis and electrophysiology, we define how specific headgroup phosphates, acyl-chain carbons, and surrounding charged residues cooperate to gate the channel. Together, these results support a model in which PIP₂ and Ca²⁺ act through a shared allosteric electrostatic network centered on the α4 helix that links lipid binding to pore opening.

Our simulations, performed in a conductance-permissive docking model, revealed that the PIP₂ headgroup engages residues on α4 and α6 in a Ca²⁺-dependent manner. In the basal state, the P1– and P5–phosphates contact basic residues on α6. Ca²⁺ binding reorients the pocket so that the P4-phosphate interacts with α4, displacing the N-terminal region of α4 and widening the extracellular pore entrance. Within this same region, carbons 6 to 15 of the acyl tails of PIP₂ form persistent van der Waals contacts with residues N546–I551 on α4. Alanine substitutions that reduce the hydrophobicity of this segment, particularly I550A, impair channel activation. Notably, TMEM16A-I550A currents are only partially rescued by short-chain diC8-PIP₂ and fully restored by long-chain PIP₂ consistent with stabilization through extended hydrophobic contacts. Consistent with these observations, TMEM16A-I550A displayed a fivefold lower apparent affinity for diC8-PIP₂ (EC₅₀ = 11 µM) compared to WT channels (EC₅₀ = 2.3 µM), underscoring the energetic contribution of acyl-chain engagement to high-affinity binding and gating. Together with earlier work identifying I550 and I551 as components of the hydrophobic gate that influence TMEM16A open probability, our data indicate that this same TMD4 segment also provides a platform for PIP₂ acyl tails to stabilize the open conformation, giving these residues a dual role in structural gating and lipid-dependent modulation.

Consistent with this framework, site-1 mutants retain Cl⁻ selectivity and rectification similar to WT TMEM16A and exhibit comparable surface expression by TIRF microscopy (SI Appendix, Figs. S4 and S5), indicating that their altered macroscopic currents are not due to major changes in ion permeation or channel abundance but instead reflect strengthened PIP₂-dependent stabilization of the Ca²⁺-activated open state.

Immediately below this hydrophobic stretch lies D554, which, together with K588 (α5), K645 (α6), and E705 (α7), forms an electrostatic gating ring at the narrowest region of the permeation pathway. These residues constitute a set of complementary charges whose orientation and spacing change upon Ca²⁺ binding. In simulations, Ca²⁺ coordination by E705 drives outward rotation of K645, expanding the pore diameter approximately twofold to threefold and enabling Cl⁻ passage. Mutational analysis confirmed that preserving the native charges maintains channel activity, whereas reversing or neutralizing them markedly reduces current without altering surface expression.

These results align with and extend previous cryo-EM and functional studies that identified K588, K645, and E705 as residues shaping the TMEM16A pore (30). Our data add D554 on α4 as a component not implicated by prior studies that completes this charge network and connects the PIP₂-responsive α4 helix to the Ca²⁺-binding α7 helix. Thus, PIP₂ binding does not act independently but instead modulates the same electrostatic ring that underlies Ca²⁺-dependent gating. This integrated mechanism unifies prior models of lipid- and Ca²⁺-mediated activation within a single structural framework.

Previous studies have proposed multiple PIP₂ interaction sites on TMEM16A, many of which overlap with the regions we designate as sites 2 to 4 in our docking analysis (10, 11, 16, 17). In our simulations, PIP₂ can bind stably at these locations, and prior mutagenesis in these regions alters channel activity, which may reflect direct lipid interactions, indirect structural perturbation, or both. Under the specific conditions used here (single PIP₂ per subunit, defined lipid composition, 1 to 2 μs conductance-permissive trajectories), however, site-2 to 4 configurations did not support sustained Cl⁻ permeation. In contrast, docking PIP₂ at the α4 to α6 interface (site 1) produced a configuration that yielded robust Cl⁻ flux in the presence of Ca²⁺ and recapitulated the characteristic headgroup and acyl-chain dependence observed in our optogenetic phosphatase, excised-patch, and chain-length experiments. Along α4, this pocket forms a continuous coupling axis in which R562 and K566 coordinate the PIP₂ headgroup, D554 contributes to the electrostatic gating ring, and N546-I551 engage the acyl tails, resulting in increased flexibility of the extracellular side of α4. On this basis, we propose that site 1 acts as a major PIP₂-dependent activation locus under our experimental and simulation conditions, while acknowledging that additional PIP₂ interactions at sites 2 to 4 or at other membrane-facing interfaces may contribute to TMEM16A regulation in a context-dependent manner. In this framework, we use allosteric to indicate that PIP₂ binds at a site spatially distinct from the Ca²⁺-binding pockets and the narrowest region of the pore yet modulates the same electrostatic gating ring that controls opening, rather than functioning as an independent orthosteric gate.

PIP₂ is the most abundant phosphoinositide in the plasma membrane and is continuously regenerated through phosphorylation of PI(4)P by PI(4)P-5-kinases (32). Although PI(4)P is comparatively minor, our findings reveal that it can support TMEM16A activity. This functionality was demonstrated both by CRY2-5P_OCRL-mediated conversion of PIP₂ to PI(4)P and by exogenous diC8-PI(4)P application to PIP₂-depleted cells. The phosphate at position 5 (P5) of PIP₂ appears dispensable for Ca²⁺-dependent gating of WT TMEM16A, whereas the phosphate at position 4 (P4) interacts robustly with residues R562 and K566 on α4. Physiologically, PIP₂ depletion during Gαq-PLC signaling leads to rapid current decay, yet the partial support provided by PI(4)P implies that the kinetics of TMEM16A activation and deactivation depend on the relative abundance of these lipids, which varies across cell types (27, 33, 34).

More generally, increased macroscopic current could arise from changes in channel number, ion permeation, or gating. Given that the surface abundance of K566A, K566Q, and the other site-1 mutants is comparable to WT TMEM16A by TIRF (SI Appendix, Fig. S4) and that their rectification and reversal potentials are similar by I-V analysis (SI Appendix, Fig. S5), our data argue most strongly for a gating-based mechanism. At the same time, K566Q exhibits a markedly higher apparent affinity for diC8-PIP₂ in excised patches (Fig. 2D) and a reconfigured headgroup contact network in our GMD simulations (SI Appendix, Fig. S6 F–H). We therefore interpret the larger whole-cell currents as reflecting stronger PIP₂-dependent stabilization of the Ca²⁺-activated open state, while acknowledging that, given TMEM16A’s very small single-channel conductance in the low-picosiemens range, we cannot fully separate effects on open probability from subtle changes in unitary conductance.

During simulations, Cl⁻ conduction occurred through one subunit at a time, suggesting possible subunit independence reminiscent of the “double-barrel” configuration of CLC chloride channels (35, 36). Determining whether TMEM16A subunits gate cooperatively or independently will require future experimental and computational work.

The PIP₂-binding pocket described here shares features common to phosphoinositide-binding domains across membrane proteins. As in Kir2.1 and Kir6.2, clusters of lysine and arginine residues at the cytoplasmic–membrane interface coordinate the PIP₂ headgroup (37–39). Many PIP₂-binding modules include at least two basic residues, an aromatic residue, and a flexible loop that reorients upon ligand binding (40, 42, 43). The TMEM16A α4–α6 interface conforms to these criteria but adds a distinctive hydrophobic element contributed by the acyl-chain interactions. This combination of electrostatic and hydrophobic coupling distinguishes the TMEM16A site from canonical PH-domain-like PIP₂ pockets while highlighting a broader mechanistic principle of membrane-anchored phosphoinositide regulation.

In summary, our results support a refined model in which PIP₂ and Ca²⁺ jointly regulate TMEM16A through an allosterically coupled electrostatic ring centered on α4. This framework reconciles lipid- and Ca²⁺-dependent gating, integrates structural and functional data across prior models, and provides a foundation for probing TMEM16A modulation and structure–activity relationships in future work.

Methods

Molecular Modeling and Docking.

The cryo-EM structure of mouse TMEM16A (PDBID: 5OYB) was prepared by the Protein Preparation Wizard module of Maestro (2019–3) program (Schrödinger, Inc., New York, NY) and subsequently utilized for docking. To systemically search the potential PIP₂-binding sites, we covered the channel with 56 grid boxes (each with centers spaced 20 Å apart) using an in-house script. Induced-fit-docking (IFD) of the Glide program from Schrödinger (Schrödinger, Inc., New York, NY) was used for docking simulations (44, 45). The diC1 PIP₂ was initially put at the center of the boxes to conduct each docking simulation, using the standard precision (SP) module in the Glide program (46). The default parameters were used for IFD simulations. The residues within 3.5 Å of ligand poses were selected for side chain optimization by prime refinement. The SP score was used for ranking the ligand poses. The pose of docked ligand with the lowest docking SP score was selected for each potential binding site. Four predicted binding sites with the lowest docking SP scores were selected, and the diC1 PIP₂ was replaced with a full-length PIP₂ for the MD simulations.

System Setup and Molecular Dynamics Simulations.

Protonation states of the titratable residues in the TMEM16 A channel were calculated at pH = 7.4 via the use of the H++ server (http://biophysics.cs.vt.edu/) (47). The simulation system was generated using the CHARMM-GUI Membrane Builder webserver (http://www.charmmgui.org/?doc=input/membrane) (48). First, the protein was inserted in an explicit lipid bilayer of POPC, POPE, POPS, and cholesterol with a molecular ratio of 25:5:5:1 (49). Then, the complexes were put in a water box (176 Å × 176 Å × 156 Å), followed by an addition of 150 mM KCl to the system.

Molecular dynamics simulations were conducted using the PMEMD.CUDA program in AMBER16. The tleap module was employed to neutralize the system by extra K⁺ or Cl⁻ ions as needed. The FF14SB, LIPID17, and GAFF force fields were chosen for protein, mixed lipid membrane, and PIP₂, respectively. The parameters of PIP₂ were generated using the general AMBER force field by the Antechamber module of AmberTools 17 and using the partial charge determined via restrained electrostatic potential charge-fitting scheme by ab initio quantum chemistry at the HF/6-31G* level. The systems were energetically minimized with 2,000 steps of steepest descent followed by 3,000 steps of conjugate gradient. Subsequently, Langevin dynamics was used to heat the systems from an initial temperature of 0 K to a final temperature of 303 K, with a collision frequency of 1 ps⁻¹. During heating, an initial constant force of 500 kcal mol⁻¹ Å⁻² was applied to positionally restrain the receptor complexes, which was then gradually reduced to 10 kcal mol⁻¹ Å⁻², allowing the lipid and water molecules to move freely. Subsequently, the systems underwent 5 ns of equilibrium molecular dynamics simulations. Finally, a total of 2 μs of production molecular dynamics simulations were conducted, with coordinates saved every 100 ps for subsequent analysis. The simulations were conducted with periodic boundary conditions, initially in a constant temperature, constant-pressure ensemble (NPT), followed by a constant-temperature, constant volume ensemble (NVT). The pressure was regulated using the isotropic position scaling algorithm, with the pressure relaxation time set to 2.0 ps. The particle mesh Ewald (PME) method was used to calculate the long-range electrostatics with a 10 Å cut-off (50). An external voltage of 0.06 V/nm (approximately 200 mV across the membrane) was added to the systems from the extracellular to the intracellular side; the treatment of the electric field has been detailed in previous work (26, 51, 53, 54). A 4-fs time step was employed using the hydrogen mass repartition algorithm for systems to accelerate the molecular dynamics simulations (55).

Analysis of MD Runs.

Distance analysis and PCA were performed using GROMACS 5.1.4 (56). PCA was performed to extract the collective motions of the channel from the MD simulation trajectories. This describes the motions with a set of eigenvector and eigenvalue pairs, which are obtained by diagonalizing the covariance matrix of the Cα atomic positional fluctuations (57, 58). The analysis program gmx_anaeig within GROMACS was used to conduct PCA. The interaction between TMEM16A and PIP₂ was analyzed by Discovery Studio 2017 software. VMD 1.9.3 was used for structure rendering (59).

Molecular Biology and Reagents.

Purified diC8-PIP₂, PI(4)P, and PI(5)P and long-chain (LC) PIP₂ were purchased from Avanti Biosciences. The LC-PIP₂ (1-stearoyl-2-arachidonoyl-sn-glycero-3-phospho-(1’-myo-inositol-4’,5’-bisphosphate) is a synthetic PI(4,5)P₂ with full-length acyl chains.

All PIP₂ variants used in the patch-clamp studies were stored in sealed glass vials with low-binding caps and kept at −20 °C when not in use. CIBN-CAAX and mCherry-CRY2-5P_OCRL were gifts from the De Camilli lab (Yale University, New Haven, CT). The open reading frame of pseudojanin (27) was designed with an N-terminal CRY2-tag in the pMAX(+) vector and was generated by Genscript (Piscataway, NY). Mouse TMEM16A (GenBank: NM_178642.6) tagged with eGFP in the peGFP-N1 expression vector was a gift from Lily Jan (UCSF). Mutations were introduced by QuikChange mutagenesis according to the manufacturer’s protocols.

Cell Culture and Transfection.

Human embryonic kidney (HEK293T) cells were acquired from American Type Culture Collection (ATCC, Cat #CRL-3216) and were maintained in Dulbecco’s modified Eagle’s medium (ATCC) supplemented with 100 units/mL penicillin, 100 μg/mL streptomycin, and 10% (vol/vol) fetal bovine serum. Cells were routinely confirmed to be Mycoplasma free via PCR and Hoechst stain. The cells were incubated in a 37 °C, humified incubator supplemented with 5% CO₂. For our experiments, cells were seeded on glass coverslips in 35 mm culture dishes at least 1 d before transfection. Cells were transiently transfected with TMEM16 A variants, CIBN-CAAX and CRY2-5P_OCRL or CRY2-pseudojanin as indicated, in OptiMEM using polyethyleneimine (PEI) for 1 to 2 h at a ratio of 1 μg of DNA to 4 μL PEI. The near-red fluorescent protein, mCherry, was typically used as a transfection marker. The cells were studied 24 to 48 h after transfection.

Patch Clamp Recording.

HEK293T cells were studied by whole-cell patch clamp. Currents were recorded with a Tecella Pico-2 amplifier (Tecella) controlled using WinWCP software (University of Strathclyde). Currents were low-pass (Bessel) filtered at 9 kHz and digitized at 50 kHz. Cells were studied in an external recording buffer comprising (in mM): 150 NaCl, 1 CaCl₂, 1 MgCl₂, 10 glucose, and 10 HEPES, adjusted to pH 7.4 with NaOH. Patch pipettes were pulled from borosilicate glass (Clark Kent), using a vertical puller (Narishige) and had a resistance of 2.5 to 4 MΩ when filled with internal solution containing (in mM): 130 CsCl, 1 MgCl₂, and 10 EGTA, with 3 Na₂ATP adjusted to pH 7.35 with CsOH. The recording conditions are based on those described by the Hille and Suh labs and give a reversal potential E_Cl of −4 mV (12). In accord with this study, we determined that 8.47 mM Ca²⁺ from CaCl₂ leaves a free Ca²⁺ concentration of 455 nΜ in the internal solution in the presence of 10 mM EGTA, under our experimental conditions, allowing robust activation of TMEM16 A channels. The free Ca²⁺ concentration is calculated using the MAXCHELATOR online algorithm housed at UC Davis. For whole-cell studies, the cells are held at −80 mV for 2 s before stepping to a 500 ms test potential between −100 mV and 100 mV with 20 mV increments. If the current-voltage (I-V) relationship is observed to deviate from a reversal potential of ~−4 mV this would indicate a change in the selectivity of the channel. For time course studies, this protocol is modified to step to 100 mV only. Whole-cell currents were normalized to cell capacitance. Mean ± SEM capacitance values were 12 ± 4 pF for HEK293 T cells.

For excised (off-cell), inside-out patch experiments, the pipettes were filled with the external recording buffer (in mM: 150 NaCl, 1 CaCl₂, 1 MgCl₂, 10 glucose, and 10 HEPES, adjusted to pH 7.4 with NaOH) and the inside face of the patches was perfused with the internal solution (in mM: 130 CsCl, 1 MgCl₂, and 10 EGTA, with 3 Na₂ATP adjusted to pH 7.35 with CsOH), containing a free Ca²⁺ concentration of 455 nM, as described above. The polarity of the voltage-clamp protocols was inverted. To study concentration-dependent PIP₂-rescue of channel activity, the current magnitude in the inside-out patch was measured then the patch was perfused with poly-L-lysine (100 µg/mL) to chelate phosphoinositides from the membrane. PIP₂ variants (diC8- or LC-PIP₂) were prepared in the ‘internal’ recording buffer and applied to the inside face of the patch using a separate, vacuum-driven delivery system that terminates in a glass pipette placed in the recording chamber close to the patch. Although we use low binding material tubing to deliver LC-PIP₂ it is possible that the final concentration of the reagent at the patch is lower than the nominal concentration. Our readout for partition of the phospholipid into the membrane was a change in function, i.e., a change in the channel current. When using mutants with reduced or eliminated channel activity, we run parallel experiments with the wild-type channel to confirm that the reagent is active.

For simultaneous optogenetic-patch clamp studies, the blue light system was photoactivated in epifluorescence-mode using continuous excitation from a broad-spectrum LED (Excelitas) through a 488/20 nm filter (Chroma), via a 20× objective lens (Olympus). The light output at the sample was measured at 50 mW/cm² by a photometer (ThorLabs). To avoid preactivation of CRY2-fusion proteins, cells were incubated in the dark and handled in foil-wrapped dishes prior to the experiments. Cells of interest were identified using mCherry as a transfection marker because its spectral properties do not activate CRY2. Further, we use a transilluminator that includes a bandpass filter to block blue light during brightfield visualization. To generate paired data, the currents were recorded first in the dark and again following a 3-min preillumination with epifluorescent blue-light that persisted through the remainder of the study. All the experiments were performed at room temperature.

TIRF Microscopy.

TMEM16A constructs, tagged at the C terminus with eGFP, were imaged at the surface of live HEK-293T cells by TIRF microscopy. Cells were seeded to #1.5 glass coverslips, transfected as described above, and studied after 24 to 32 h in a bath solution comprising (in mM): 150 NaCl, 1 CaCl₂, 1 MgCl₂, 10 glucose, and 10 HEPES, adjusted to pH 7.4 with NaOH. The eGFP-tag was excited by a 488-nm laser (Coherent, Santa Clara, CA). The beam was conditioned for coherence using a custom-built Keplerian beam expander upstream of a 488/10 nm laser cleanup filter. The laser line was tuned to provide 10 mW of incident light on a micro mirror positioned below a high numerical-aperture apochromat objective (60×, 1.5 NA; Olympus, Waltham, MA) mounted on an RM21 microscope frame equipped with a piezo-driven nanopositioning stage (MadCity Labs., Madison, WI). The emission of eGFP was isolated from the excitation beam by an exit micro mirror and a ring-diaphragm positioned below the micro mirror assembly. The emission was imaged using a back illuminated sCMOS camera (Teledyne Photometrics, Tucson, AZ) controlled by Micro-Manager freeware (UCSF). All filters and mirrors were from Chroma (Bellow Falls, VT). Lenses, pinholes, and diaphragms were from ThorLabs (Newton, NJ). TetraSpeck beads (Thermo, Waltham, MA) were routinely imaged to map the sCMOS chip and to calibrate the evanescent field depth to 100 nm (60). For each construct, images were acquired on at least three independent experimental days, with 3 to 4 coverslips analyzed per day. On each coverslip, 10 to 20 nonoverlapping fields were imaged, and rectangular regions of interest (~0.5 to 1 µm²) were drawn over the ventral membrane of individual cells. Laser power, exposure time, camera gain, and the 100 nm evanescent field depth were kept constant across all constructs within an experiment. For each construct, 30 to 60 cells were quantified in total. The fluorescence intensity of cells was determined in ImageJ, with untransfected cells used to determine the background level of fluorescence. Control experiments were performed using eGFP-tagged GIRK potassium channels. GIRK1-eGFP has limited expression at the cell membrane when expressed alone and was used as a negative control (61). GIRK2 is untagged. GIRK1-eGFP forms heterotetrametric channels upon coexpressed with untagged GIRK2 (62). This condition was a positive control for GIRK1-eGFP fluorescence.

Statistics.

Data were handled in WinWCP, Clampfit, and Excel software with statistical analysis performed using GraphPad (Prism). The data are presented as mean ± SD with statistical differences between paired groups determined by two-tailed, unpaired, Student’s t test, unless indicated otherwise. The threshold for significance was determined to be P < 0.05.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All input files, initial structures, and final coordinate outputs associated with the TMEM16A-PIP2 system induced-fit docking (IFD) and molecular dynamics (MD) simulations have been deposited in the Zenodo public repository under accession link https://doi.org/10.5281/zenodo.18361676 (19). All other data are included in the article and/or SI Appendix.