Since the elegant experiments that led to the discoveries of TMEM16A/ANO1 and TMEM16B/ANO2 as the long-sought-after Ca²⁺-activated Cl⁻ channels (CaCCs) in 2008 (Caputo et al., 2008; Schroeder et al., 2008; Yang et al., 2008), substantial progress has been made to understand the biology of this unique family of transmembrane proteins. Numerous studies confirmed that TMEM16A and TMEM16B are responsible for the endogenous CaCC currents observed in various cell types (Bader et al., 1982; Miledi, 1982; Barish, 1983). More excitingly, new findings uncovered their novel physiological and pathological functions, including smooth muscle contraction, trans-epithelial fluid transport, secretion, tumor progression, sensory transduction, mood control, and motor learning (Hartzell et al., 2005; Duran and Hartzell, 2011; Pedemonte and Galietta, 2014; Oh and Jung, 2016; Whitlock and Hartzell, 2016a; Zhang et al., 2017; Crottes and Jan, 2019).

Among the most striking findings in TMEM16 research is that, unlike initial predictions, the remaining family members are likely not CaCCs. Instead, the majority of the TMEM16 family members characterized thus far are Ca²⁺-activated phospholipid scramblases (CaPLSases), which can translocate phospholipids down their chemical gradients in a relatively non-selective fashion. As passive phospholipid transporters, TMEM16 CaPLSases can efficiently translocate phospholipids at high speed (4.5×10⁴ phospholipids per second for TMEM16F; Watanabe et al., 2018). Therefore, activation of TMEM16 CaPLSases leads to rapid collapse of membrane phospholipid asymmetry, which can trigger a plethora of cellular responses and physiological functions, such as blood coagulation (Suzuki et al., 2010; Yang et al., 2012), microparticle release (Fujii et al., 2015), membrane repair (Wu et al., 2020), sheddase activation (Sommer et al., 2016; Veit et al., 2018; Bleibaum et al., 2019), endosomal sorting (Petkovic et al., 2020), cell–cell fusion (Griffin et al., 2016; Whitlock et al., 2018; Zhang et al., 2020; Braga et al., 2021), and viral infection (Bevers and Williamson, 2016; Zaitseva et al., 2017; Younan et al., 2018). While the list of new biological functions of TMEM16 CaPLSases and CaCCs keeps growing, their importance in human health and disease has become apparent, as malfunctions in TMEM16 proteins have been implicated in human diseases, including asthma, cancer, bleeding disorders, muscular dystrophy, arthritis, epilepsy, dystonia, and ataxia (Duran and Hartzell, 2011; Pedemonte and Galietta, 2014; Oh and Jung, 2016; Crottes and Jan, 2019). To target TMEM16 proteins and treat TMEM16-related diseases, it is critical to have a comprehensive understanding of these novel proteins at the molecular level.

Structural, functional, and computational characterizations of TMEM16 proteins have provided an in-depth understanding of the mechanisms of permeation, activation, and regulation. Given the space limit of this review, we first briefly summarize the key structural features of TMEM16F CaCCs and CaPLSases and then focus on discussing the molecular mechanism of Ca²⁺-dependent gating, and how an allosteric Ca²⁺ binding site, phosphatidylinositol-(4,5)-bisphosphate [or PI(4,5)P₂], and pH regulate TMEM16 Ca²⁺-dependent gating. This is by no means a comprehensive review of TMEM16 structure and function. The readers are encouraged to refer to the excellent reviews of the biophysics (Brunner et al., 2016; Whitlock and Hartzell, 2016b; Falzone et al., 2018; Kalienkova et al., 2021; Le and Yang, 2021) and physiology of TMEM16 proteins (Hartzell et al., 2005; Duran and Hartzell, 2011; Pedemonte and Galietta, 2014; Oh and Jung, 2016; Whitlock and Hartzell, 2016a).

The first glimpse into the atomic structure of TMEM16 proteins came from the X-ray structures of a fungal TMEM16 homolog from Nectria haematococca (or nhTMEM16, Figure 1A Left; Brunner et al., 2014), which functions as a CaPLSase and likely also a Ca²⁺-activated nonselective channel (Lee et al., 2016). Subsequent structural analyses of the fungal afTMEM16, mouse TMEM16A, mouse TMEM16F, and human TMEM16K all revealed their highly conserved architecture (Brunner et al., 2014; Dang et al., 2017; Paulino et al., 2017a,b; Alvadia et al., 2019; Falzone et al., 2019; Feng et al., 2019; Kalienkova et al., 2019). Similar to ClC Cl⁻ channels and Cl⁻/H⁺ exchangers (Miller, 2006), a functional TMEM16 protein is a dimer with a double-barreled architecture, in which an independent permeation pore resides in each subunit. The double-barreled architecture was functionally validated by electrophysiological characterizations of TMEM16A concatemers, where each monomer possessed different Ca²⁺ sensitivities or ion selectivities (Jeng et al., 2016; Lim et al., 2016).

Different from the initial prediction of an 8-transmembrane (TM) topology, we now know that each TMEM16 monomer consists of 10 TM segments preceded by a long N-terminal cytosolic domain (NCD) and followed by a short C-terminal extension of TM10 (Figure 1B). TM7 and TM8 do not completely traverse the membrane, which, together with TM6, form two highly conserved Ca²⁺ binding sites (Figures 1B, 2). The anion permeation pathway of the TMEM16A is shaped like an asymmetric hourglass and is formed by numerous hydrophilic and nonpolar residues from TMs 3–7. The so-called hydrophilic cavity has been shown to form a non-selective permeation pathway for not only ions in the TMEM16 channels, but also phospholipids in the scramblases (Brunner et al., 2014; Dang et al., 2017; Jiang et al., 2017; Paulino et al., 2017a,b; Lee et al., 2018; Bushell et al., 2019; Falzone et al., 2019; Le et al., 2019b). Notably, in the fungal nhTMEM16 and afTMEM16 as well as the human TMEM16K structures, the hydrophilic cavity has been captured in an “open” conformation in which the peripheral TM4 and TM6 are physically separated, exposing the hydrophilic cavity to the lipid environment (Figure 1A Right; Falzone et al., 2018; Kalienkova et al., 2021). This putative “open” lipid-conducting state supports the notion that TMEM16 scramblases catalyze lipid translocation via a “credit card” model previously proposed for phospholipid flippases (Pomorski and Menon, 2006). This model implies that the headgroups of permeating phospholipids may slide along the hydrophilic groove of TMEM16 scramblases, while their acyl tails remain in the hydrophobic lipid environment, a hypothesis that has been supported by extensive structural, functional, and molecular dynamics (MD) studies (Brunner et al., 2014; Bethel and Grabe, 2016; Jiang et al., 2017; Lee et al., 2018; Bushell et al., 2019; Kalienkova et al., 2019; Le et al., 2019b). For dual function ion channel/scramblases, ions may permeate adjacent to lipid headgroups through a proteolipid pore (Whitlock and Hartzell, 2016b). In support of this idea, a recent computational study suggested that the ion permeation pathway in the fungal nhTMEM16 and human TMEM16K is partially lined by ordered lipid headgroups (Kostritskii and Machtens, 2021). The lipid headgroup identity, pore-lining residues, and membrane voltage all exert appreciable effects on ion permeation and selectivity (Kostritskii and Machtens, 2021). By contrast, all current Ca²⁺-bound structures of the TMEM16A CaCC and the dual function TMEM16F ion channel/scramblase paradoxically adopt tightly closed permeation pathways that are too narrow to allow the passage of ions or lipids (Dang et al., 2017; Paulino et al., 2017a; Alvadia et al., 2019; Feng et al., 2019; Figures 2A,B). The reason for these structural observations remains elusive and requires future investigation.

Based on structural, functional, and computational evidence of Ca²⁺-dependent activation and PI(4,5)P₂-dependent regulation (see in the next sections), we recently proposed a modular model of TMEM16 proteins to simplify the complex TMEM16 architecture (Figures 1B–D; Le et al., 2019a). According to this model, a TMEM16 monomer can be divided into several structurally and functionally distinct domains and modules. Besides the NCD, the transmembrane region can be divided into two domains: the pore-gate domain (PGD) and the supporting domain. Consisting of TMs 3–8, the PGD not only forms the permeation pathway for ions and phospholipids, but also harbors the highly conserved primary Ca²⁺ binding sites (Figures 2A,C) and, in the case of TMEM16A, a putative PI(4,5)P₂ binding site (Figures 1B,C). As Ca²⁺ binding and PI(4,5)P₂ binding are structurally and functionally segregated, the PGD of TMEM16A can be further divided into two modules (Le et al., 2019a). The Ca²⁺ module consists of TMs 6–8 and is responsible for binding Ca²⁺ and initiating Ca²⁺-dependent activation. The regulatory module (TMs 3–5) forms the other half of the PGD, which works synergistically with the Ca²⁺ module to facilitate TMEM16 gating and permeation. In TMEM16A, PI(4,5)P₂ binding to the regulatory module stabilizes the open state and prevents the PGD from collapsing and entering the desensitized state. TMs 1, 2, 9, and 10, on the other hand, form the supporting domain. As will be discussed below, the supporting domain contains a conserved third Ca²⁺ binding site that is allosterically coupled to the PGD (Figure 1B). The supporting domain also helps establish the dimer interface within the membrane through inter-subunit interactions between the extracellular regions of TM10. The dimer interface creates two large hydrophobic cavities, or dimer cavities, along the central axis of all TMEM16 proteins. In addition to the TM10 interactions, the fungal nhTMEM16 and afTMEM16 as well as the human TMEM16K adopt a “domain-swapped” organization in which their C-terminal ends have extensive inter-subunit interactions with the NCD of the adjacent protomer. Interestingly, “domain-swapping” is not observed in the TMEM16A and TMEM16 F structures. Beyond facilitating dimer formation, supporting domain interactions potentially serve to stabilize the PGD.

It is worth noting that four conserved disulfide bonds constrain the relatively long extracellular loops connecting TM1-2, 3–4, 5–6, and 9–10 in most mammalian TMEM16 proteins, including TMEM16A and TMEM16F. Disruption of these disulfide bonds leads to dysfunctional channels (Yu et al., 2012), suggesting that the stability of the extracellular loops is important for TMEM16A and TMEM16F activity. Interestingly, the extracellular loops of the fungal nhTMEM16 and afTMEM16 and the endoplasmic reticulum (ER)-resident human TMEM16K scramblase are significantly shorter and lack disulfide bonds (Brunner et al., 2014; Bushell et al., 2019; Falzone et al., 2019). Since these TMEM16 proteins still function as CaPLSases and Ca²⁺-activated nonselective ion channels, the long extracellular loops seem dispensable for ion and lipid transport functions. The precise functions of the extracellular loops are unclear and require future investigation.

In summary, structure/function studies have elucidated many of the defining elements of the TMEM16 family. These elements may be conserved in the evolutionarily related osmo- and mechano-sensing OSCA/TMEM63A (Murthy et al., 2018) and TMC channels (Hahn et al., 2009; Ballesteros et al., 2018; Medrano-Soto et al., 2018). To simplify the growing complexity, we propose a modular design of the TMEM16 proteins (Figures 1B,C), which may also facilitate the understanding of the molecular mechanisms of related proteins.

All mammalian TMEM16 ion channels and lipid scramblases require the binding of intracellular Ca²⁺ for activation, albeit at different levels of potency. The TMEM16A and TMEM16B CaCCs are synergistically gated by intracellular Ca²⁺ and membrane voltages. TMEM16A is highly sensitive to Ca²⁺ with an estimated EC₅₀ of 0.4 to 1μM at positive membrane potentials or 0.7 to 6μM at negative membrane potentials (Yang et al., 2008; Ferrera et al., 2009; Xiao et al., 2011; Yu et al., 2012; Brunner et al., 2014; Ni et al., 2014; Tien et al., 2014; Lim et al., 2016; Cruz-Rangel et al., 2017; Le et al., 2019a). Despite their similarity, TMEM16B displays a lower Ca²⁺ sensitivity with an estimated EC₅₀ of around 1.2 to 3.3μM at positive membrane potentials and 1.8 to 4.9μM at negative potentials (Pifferi et al., 2009; Stephan et al., 2009; Cenedese et al., 2012; Adomaviciene et al., 2013; Pifferi, 2017). One notable feature is that TMEM16A- and TMEM16B-mediated currents are outward rectifying under the low open probability and display time-dependent activation and deactivation kinetics (Caputo et al., 2008; Yang et al., 2008; Pifferi et al., 2009; Stephan et al., 2009). However, these channels are no longer time- and voltage-dependent when they are fully opened by saturating Ca²⁺.

The dual function TMEM16F ion channel and phospholipid scramblase is less sensitive to Ca²⁺. The estimated EC₅₀ values range from 3.4 to 105μM, depending on the configuration and ionic conditions of the patch clamp recording (Yang et al., 2012; Grubb et al., 2013; Shimizu et al., 2013; Scudieri et al., 2015; Feng et al., 2019; Le et al., 2019b; Nguyen et al., 2019; Ye et al., 2019). The Ca²⁺ sensitivity for TMEM16F scrambling activity has not been accurately measured. However, based on the co-occurrence of TMEM16F current and scramblase activity recorded using patch clamp-lipid scramblase fluorometry (PCLSF) assay (Yu et al., 2015; Liang and Yang, 2021), it is expected that the Ca²⁺ sensitivity for TMEM16F CaPLSase activity is comparable to the Ca²⁺ sensitivity for channel activity. TMEM16F-mediated ionic conductance is elicited by the synergistic activation of membrane depolarization and Ca²⁺ binding (Yang et al., 2012). Unlike TMEM16A and TMEM16B, the TMEM16F channel always requires membrane depolarization for activation and its current remains strongly outward rectifying even at high Ca²⁺ concentrations. It is yet unknown whether membrane voltage can promote CaPLSase activity.

Mutagenesis studies on TMEM16A CaCC successfully identified five highly conserved acidic residues as putative Ca²⁺ binding residues, including E650 on TM6, E698 and E701 on TM7, E730 and D734 on TM8 (Yu et al., 2012; Tien et al., 2014; Figures 2A,C, numbering based on the TMEM16A(a) isoform lacking the EAVK segment). Neutralizing mutations (to alanine or glutamine) strongly reduce the Ca²⁺ sensitivity of TMEM16A from the sub-micromolar range to the millimolar range. Subsequent structural and functional studies not only validated these electrophysiological findings but also revealed three additional asparagine residues (N646 and N647 of TM6 and N726 of TM8) as additional Ca²⁺ coordinates (Brunner et al., 2014; Dang et al., 2017; Paulino et al., 2017a). Within each TMEM16 monomer, the Ca²⁺ binding residues cluster together and form two highly conserved Ca²⁺ binding sites, herein referred to as the primary Ca²⁺ sites (Figures 1B, 2). The highly conserved primary Ca²⁺ binding sites among different TMEM16 homologs suggest that these evolutionarily conserved proteins maintain a similar activation mechanism.

The primary TMEM16 Ca²⁺ binding sites have several unique features (Figures 1B, 2). First, the Ca²⁺ binding residues reside within the membrane electrical field, which is in excellent agreement with a previous prediction (Arreola et al., 1996). The membrane location of the Ca²⁺ binding sites in TMEM16 proteins may partially contribute to their weak voltage-dependent Ca²⁺ activation (Hartzell et al., 2005; Pifferi et al., 2009; Xiao et al., 2011; Yang et al., 2012), as Ca²⁺ ions need to travel within the membrane electric field to reach the binding sites. Second, the primary Ca²⁺ binding sites are located near the ion/lipid permeation pathway. Such proximity between the Ca²⁺ binding sites and the activation gates implies that TMEM16 proteins can efficiently transmit Ca²⁺ binding energy to operate their activation gates.

Structural and functional studies have shown that Ca²⁺-induced TM6 conformational changes are critical for Ca²⁺-dependent activation of both TMEM16 ion channels and scramblases (Figures 1, 2; Dang et al., 2017; Paulino et al., 2017a; Peters et al., 2018; Alvadia et al., 2019; Feng et al., 2019). Structural studies of the TMEM16A CaCC showed that in the absence of Ca²⁺, TM6 adopts an alpha-helical conformation with a kink at G640 (Figures 2A,C; Dang et al., 2017; Paulino et al., 2017a). This kink causes the C-terminal segment of TM6 to swing away from TM7 and TM8, thereby rendering the negatively charged Ca²⁺ binding residues accessible to the cytosol. The highly electronegative environment created by the apo Ca²⁺ binding sites also serves to impede Cl⁻ entry from the intracellular side (Paulino et al., 2017b; Lam and Dutzler, 2018). It was suggested that Ca²⁺ ions first bind to and neutralize the four highly acidic residues from TM7 and TM8, providing an attractive environment that allows TM6 to move toward TM7 and TM8 by interacting with the bound Ca²⁺ ions via N647 and E650. During this process, TM6 rotates around the G640 hinge because of the interactions between N647, E650, and the two bound Ca²⁺, subsequently leading to the formation of a π-helix (Figure 2A). Superimposing the Ca²⁺-bound and Ca²⁺-free structures reveals that Ca²⁺ binding leads to partial widening of the central constriction site in TMEM16A, though, paradoxically, the permeation pathway is still too narrow for anion passage. Supporting the functional importance of TM6 in TMEM16A gating, several mutations on TM6 such as I637A, I637K, G640A/P, Q645A, and P654A were shown to alter the channel’s Ca²⁺ sensitivity (Dang et al., 2017; Paulino et al., 2017a; Lam and Dutzler, 2018; Peters et al., 2018; Le et al., 2019b). These mutations likely shift the equilibrium of TM6 to favor either the open conductive state (G640A/P, I637A/K, and Q645A) or the closed non-conductive state (P654A). A recent computational study further supports the importance of TM6 conformational changes in Ca²⁺-dependent gating of TMEM16A (Shi et al., 2021). Based on MD simulations, the authors concluded that separation of TM6 and TM4 may lead to expansion of the ion permeation pore and consequently the opening of the channel. This is consistent with the “pore-dilation” model (Figure 1C) derived from functional tests (Le et al., 2019b).

Conformational changes of TM6 also seem critical for the gating of the TMEM16F ion channel/scramblase, albeit via an opposite movement of the cytosolic end of TM6 compared to TMEM16A TM6 (Figure 2). However, analogous to TMEM16A, binding of two Ca²⁺ ions to N620, N621, and E624 of TM6, E667 and E670 of TM7, and E699 and D703 of TM8 neutralizes the Ca²⁺ binding sites and allows TM6 to approach TM7 and TM8 via a rigid body movement around G615, equivalent to TMEM16A’s G640 (Figure 2). Because of a missing residue near the G615 hinge (Figure 2C), Ca²⁺ binding does not result in partial unwinding of TM6 and hence the π-helix does not form in TMEM16F (Alvadia et al., 2019; Feng et al., 2019). A similar transition from a bent to straight conformation of TM6 was also observed in the structures of TMEM16F with zero or one Ca²⁺ bound, respectively (Feng et al., 2019). It is worth noting that while the fungal afTMEM16 and nhTMEM16 homologs lack a glycine hinge, TM6 also undergoes a similar swinging movement around the equivalent region upon Ca²⁺ binding (Falzone et al., 2019; Kalienkova et al., 2019). These observations further illuminate the conserved gating mechanism shared among TMEM16 ion channels and scramblases.

While Ca²⁺-induced conformational changes in TM6 were unambiguously shown to be critical for the gating of TMEM16 ion channels and scramblases, recent studies on TMEM16A (Tak et al., 2019) and TMEM16F (Roh et al., 2021) proposed another interesting Ca²⁺-dependent gating. Tak et al. (2019) suggested that the TMEM16A CaCC harbors an EF-hand-like domain consisting of a cluster of acidic residues (TMEM16A D285 to D297) that could serve as a reservoir for Ca²⁺ binding before being transferred to the primary sites in TMs 6–8 for subsequent activation. Neutralization of these acidic residues reduces both TMEM16A’s Ca²⁺ and voltage sensitivity. While TMEM16F does not appear to have such an EF-hand-like domain, Roh et al. (2021) showed that neutralizing acidic residues in the equivalent N-terminal domain of TMEM16F reduces its Ca²⁺ sensitivity, consistent with the importance of this acidic Ca²⁺ reservoir in channel gating. Furthermore, the N-terminal Ca²⁺ reservoir in TMEM16F has less acidic residues compared to that of TMEM16A and contains additional basic residues. Replacing the N-terminal Ca²⁺ reservoir of TMEM16F with the equivalent EF-hand-like N-terminal domain of TMEM16A markedly enhances TMEM16F’s Ca²⁺ sensitivity, suggesting that the differences in electronegativity at this region may contribute to determining Ca²⁺-dependent gating in TMEM16 proteins (Tak et al., 2019; Roh et al., 2021).

Another intriguing phenomenon about TMEM16F Ca²⁺-dependent activation is the long (~5–10min) delay after establishing the whole-cell patch clamp configuration (Grubb et al., 2013; Shimizu et al., 2013; Scudieri et al., 2015; Yu et al., 2015; Lin et al., 2018; Liang and Yang, 2021; Stabilini et al., 2021). This delay persists even when the pipette solution contains 100–200μM Ca²⁺. Therefore, the delay cannot be simply explained by the relatively low Ca²⁺ sensitivity of TMEM16F, which may require prolonged diffusion time for intracellular Ca²⁺ to reach the threshold concentration to activate TMEM16F. Paradoxically, TMEM16F current can be instantaneously activated without delay under inside-out configuration (Yang et al., 2012; Lin et al., 2018; Liang and Yang, 2021). It seems apparent that some intracellular factors might be responsible for the patch configuration-dependent discrepancy on TMEM16F activation. Although the detailed mechanisms are still unclear, a recent study provided important clues (Lin et al., 2018). The authors found that disrupting the actin cytoskeleton with cytochalasin-D (cytoD) significantly shortens the delay and accelerates TMEM16F activation. Analogously, the actin filament-stabilizing agents phalloidin and jasplakinolide inhibit TMEM16F current activation. These results suggest that the actin cytoskeleton may negatively regulate TMEM16F ion channel activity under the whole-cell configuration. Interesting, the authors also showed that intracellular magnesium ATP but not sodium ATP further prolongs the delay for TMEM16F current activation. How these intracellular factors affect TMEM16F current activation and if they also affect TMEM16F lipid scrambling activity warrant further investigations.

Structural, functional, and computational studies have demonstrated a crucial role for pore-lining TM6 residues in gating of both TMEM16 channels and TMEM16 scramblases. However, a comprehensive understanding of their gating mechanisms requires the identification of the physical activation gate that opens and closes to control ion and phospholipid permeation in response to Ca²⁺ binding. Such activation gates have been proposed for both TMEM16A and TMEM16F (Le et al., 2019b; Lam et al., 2021). Using MD simulations and an optimized lipid scrambling assay, three bulky and hydrophobic residues–F518 in TM4, Y563 in TM5, and I612 in TM6, were identified as the major constituents of the scramblase inner steric activation gate in TMEM16F (Figure 1B; Le et al., 2019b). Removing steric hindrance via alanine substitutions of these residues leads to constitutively active TMEM16F scramblases, whereas substitution with leucine or a bulky tryptophan strongly impairs TMEM16F scrambling activity following Ca²⁺ stimulation. On the other hand, mutating the inner gate with polar or charged residues greatly enhances TMEM16F lipid scrambling and ion channel activities. Most of these mutations require culturing the transfected cells in Ca²⁺-free media to suppress TMEM16F gain-of-function (GOF)-induced cytotoxicity, suggesting that basal Ca²⁺ activity is sufficient to open the inner activation gate. Remarkably, F518K and Y563K result in constitutively active TMEM16F scramblases even when the primary Ca²⁺ binding sites are destroyed. More strikingly, the TMEM16A L543K mutation, equivalent to TMEM16F F518K, converts the TMEM16A CaCC into a GOF phospholipid scramblase (Le et al., 2019b). Based on these functional observations and various TMEM16 scramblases captured in different conformations (Alvadia et al., 2019; Falzone et al., 2019; Feng et al., 2019; Kalienkova et al., 2019), a “clam-shell” model was proposed to describe the Ca²⁺-dependent gating of the TMEM16 phospholipid permeation pathway (Le et al., 2019b; Figure 1D). According to this model, Ca²⁺-induced conformational changes at the primary Ca²⁺ binding sites interrupt the interactions between TM4 and TM6 in the neck region, leading to the separation of TM4 and TM6. This clam-shell-type opening exposes the hydrophilic interior of the permeation pathway to the hydrophobic phase of the membrane, thereby allowing phospholipid headgroups to gain access and scramble (Figures 1B,D). Clam-shell opening also enables ion permeation through the proteolipid pore. Replacing the bulky, hydrophobic residues at the inner activation gate with smaller, polar, or charged amino acids weakens the interactions between TM4 and TM6, leading to enhanced permeation or a constitutively open permeation pathway for both lipids and ions.

As Cl⁻ permeation through CaCC requires an enclosed protein environment, it is conceivable that TMEM16A gating may not follow the “clam-shell” gating model of the TMEM16 scramblases. Instead, Ca²⁺-induced conformational changes only appear to dilate the central pore of TMEM16A, allowing tight control of Cl⁻ permeation (Dang et al., 2017; Paulino et al., 2017a; Le et al., 2019b; Shi et al., 2021; Figure 1C). The hydrophobic residues L543, I546, I547, and I637 (L547, I550, I551, and I641 in the (ac) isoform) at the equivalent locations to the TMEM16F inner gate residues likely form the hydrophobic gate to control TMEM16A Cl⁻ permeation (Le et al., 2019b; Lam et al., 2021) as evidenced by alanine and lysine mutations promoting TMEM16A activation. Interestingly, L543K enables TMEM16A activation in the absence of Ca²⁺ and reduces its anion selectivity, in addition to converting TMEM16A into a phospholipid scramblase as mentioned above (Le et al., 2019b). Interestingly, a previous discovery showed that substitution of a 35 amino acid segment spanning TM4 and TM5 of TMEM16A with the corresponding segment in TMEM16F rendered TMEM16A capable of scrambling phospholipids (Yu et al., 2015). Inspired by the MD simulations of fungal nhTMEM16, a follow-up study identified three additional mutations (V543S, V543T, K588N, numbering based on the TMEM16A(ac) isoform) on two pore lining residues, which can also convert TMEM16A CaCC into lipid scramblases (Jiang et al., 2017). These functional studies thus imply that TMEM16A CaCC may preserve an evolutionary potential to permeate phospholipids. The width of TM4/TM6 separation during gating is likely the key structural determinant for a TMEM16 protein to serve as a pure ion channel or a phospholipid scramblase (Figures 1C,D). For a TMEM16 CaPLSase, Ca²⁺ binding induces wide opening of the TM4/TM6 interface, thereby allowing phospholipid headgroups to gain access and scramble. On the other hand, TM4/TM6 of TMEM16 CaCCs clash with each other in the neck region of the permeation pathway, which prevents them from separating. Therefore, Ca²⁺ binding only allows ion flux without phospholipid permeation. When a charged mutation at the inner gate weakens the interactions between TM4 and TM6, the interface between the two helices may be forced to open widely so that phospholipids can permeate. Future structural, functional, and computational studies are needed to test this hypothesis. It is worth noting that endogenous CaPLSases are ubiquitously expressed in various cell lines (Kunzelmann et al., 2009). Therefore, a cell line without endogenous CaPLSase activity (Le et al., 2019b,c; Liang and Yang, 2021) is essential to experimentally examine the mutational effects on scrambling activities.

Structural, functional, and computational studies in the past decade have greatly advanced our understanding of TMEM16 proteins at the molecular level. In the next phase, the answers to the following questions will further advance our understanding of these enigmatic proteins. First, it will be important to demonstrate how the third Ca²⁺ site is allosterically coupled to the PGD and how all three Ca²⁺ bindings sites synergistically control TMEM16 activation under physiological conditions. Second, future investigations are needed to dissect how Ca²⁺ and voltage synergistically operate TMEM16 gating. The answer to this question is critical to uncover the physiological functions of TMEM16 proteins in excitable cells such as neurons and muscles. Together, we have started to understand the molecular mechanisms of TMEM16 ion and lipid permeation and identified several molecular determinants that define whether a TMEM16 protein is a sole ion channel or a dual function scramblase/ion channel. Comprehensive studies are needed to demonstrate how ion and phospholipid permeation are dynamically controlled by Ca²⁺- and voltage-induced conformational changes in the PGD. Substantial progress has been made on deciphering how pH and PI(4,5)P₂ regulate TMEM16 proteins. Identifying other physiological regulatory factors, such as post-translational modifications, are needed to further reveal how TMEM16 protein activities are fine-tuned under physiological conditions. Additionally, the Ca²⁺-bound TMEM16A and TMEM16F structures were captured in non-conductive states. Future structural studies are needed to capture the open conformations, which will enhance our understanding of TMEM16 gating transitions in response to Ca²⁺ and voltage stimulation. Apart from the ER-resident TMEM16K, the other mammalian TMEM16 proteins expressed in intracellular organelles are largely uncharacterized. Functional and structural characterization of these TMEM16 proteins will help us better evaluate their biological functions in health and disease. Finally, the evolutionary relationships between TMEM16, OSCA/TMEM63, and TMC proteins within the transmembrane channel-scramblase (TCS) superfamily are intriguing. A combination of structural, functional, and computational approaches is needed to unveil the molecular underpinnings of how this superfamily of membrane ion channels and scramblases posses different permeation, activation, and gating properties. In this review, we summarize the collective efforts from the TMEM16 field over the past decade. We propose a “modular design” model for TMEM16 assembly, and the “clam-shell” and “pore-dilation” gating/permeation models for TMEM16 scramblases and channels, respectively. We hope these simplified models serve as a steppingstone for answering the aforementioned questions, driving the field forward.

SCL, PL, AJL, and HY wrote the manuscript. SCL plotted all the figures. All authors contributed to the article and approved the submitted version.

This work was supported by NIH grant DP2-GM126898 (to HY) and American Heart Association Pre-doctoral Fellowship 19PRE34380456 (to SCL).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.