Greasing the Gears of Mechanosensitive Piezo Channels with Phosphoinositides and Cholesterol

Main Text

Ion channels are crucial for all life, enabling the transport of ions, that carry either excitatory or inhibitory stimuli across otherwise nonconductive lipid membranes (1). These channels are tightly regulated by voltage, ligands, pH, and mechanical pressure, the latter of which has come more into the focus in the recent years. These force-sensitive channels are found across the tree of life and in a variety of shapes and sizes. Mechanosensitive MscS and MscL channels in prokaryotes respond to turgor pressure to prevent cell lysis, whereas members of the K2P and TRP channel superfamilies in eukaryotes are responsive to mechanical stretching of the membrane and have roles in, for example, pain sensation. In this issue of the Biophysical Journal, Buyan et al. (2) focus on a mechanically sensitive ion channel that was first identified to be encoded by the FAM38A gene (3). In a seminal study published in 2010, this gene was shown to be required for the expression of ion channels that are activated by pressure. This resulted in the reclassification of both the gene and protein names as PIEZO1, derived from the Greek “πίεση” (pίesi), meaning pressure. The homologous gene FAM38B was also reclassified as PIEZO2. These considerable proteins, ∼2500 amino acids strong, are fundamental to the mechanosensation that underlies our sense of touch, hearing, and balance, as well as our ability to regulate blood and osmotic pressures. Genetic mutations of these channels, or channelopathies, are implicated in a number of diseases, including xerocytosis (red blood cell dehydration through excess K⁺ ion export) and lymphatic dysplasia, an incorrect distribution of lymph fluid in the body.

With the well-documented revolution in the field of cryogenic electron microscopy, this large trimeric ion channel became an attractive target for structural determination, especially after the deposition of a 4.8-Å map in 2015 (4). This prompted a race to capture higher resolution structures of PIEZO1, resulting in three almost back-to-back studies in Nature and eLife, in 2017–2018, detailing the structure of mouse PIEZO1 structure at sub-4-Å resolution (5, 6, 7). The structures revealed a triskelion arrangement of a PIEZO1 homotrimer, with a central pore lined by two transmembrane (TM) helices per subunit and surrounded by three extended propeller blades, with a single arm consisting of an individual protein subunit. A PIEZO1 monomer is predicted to be made up of as many as 38 TM helices, with the helices in the propeller blades grouped into four-helix bundles. As of yet, none of the current structures capture the first ∼500 residues of the protein, with this N-terminal domain proposed to contain at least 12 of the TM helices grouped into three four-helix bundles.

Rather than resting planar and parallel to the membrane, the TM domains in the trimeric architecture are more curved toward the intracellular side of the membrane (Fig. 1 a). This suggests that for the resolved cryogenic electron microscopy “resting-state” structures, the membrane deforms around the protein to encase the TM helices with lipids and proposes a concerted structural rearrangement within the trimeric unit to open the channel in response to mechanical pressure.

Figure 1.

Lipid interactions of mPIEZO1. (a) A coarse-grained (CG) representation of an mPIEZO1 trimer (green, blue, and yellow surface representation) within a highly-curved membrane, denoted by a dashed orange line to represent the phosphate particles of the membrane, is shown. A proposed binding site for PIPs is shown by a black box. (b) A magnification of the site, depicting lipid binding to PIEZO1 at two of the identified binding sites for PIPs, is shown here with two CG PIP₃ (purple sticks/pink spheres) bound to site 1 (¹⁶⁶KKKK²¹⁶⁹) and site 2 (K2096, K2097, and K2163). Lysines are colored blue. A CG cholesterol (yellow sticks) is also shown bound, adjacent to Y2173 (cyan; annotated Y). To see this figure in color, go online.

Given the importance of the lipids within the membrane around the structure, Buyan et al. sought to identify specific lipid binding sites for phosphoinositides and cholesterol with mouse PIEZO1 (2). This study brings together the ion channel expertise from the Corry group with the large-scale mixed membrane simulation capability of the Marrink lab and combines it with the experimental ion channel force-sensation know-how of the Martinac mechanobiologists.

The aim of the study is to capture specific lipid and sterol binding sites, to relate this to channel function, and to highlight the residues involved that have associated channelopathies. The authors initiate the study by performing coarse-grained (CG) molecular dynamics (MD) simulations of a trimer of mPIEZO1 within a lipid bilayer containing more than 60 species of lipids to best represent a complex mammalian bilayer (8). The CGMD simulations with the Martini 2.2 force field result in a binding of specific lipids within and around the annular shell of the protein. In particular, this yields annular enrichment and defined binding sites in the cytoplasmic-facing membrane leaflet, for phosphoinositides, including phosphatidylinositol-3,4-bisphosphate and phosphatidylinositol-3,4,5-trisphosphate (PIP₃). It also reveals extensive glycolipid binding in the extracellular-facing leaflet that should yield questions for further studies. From these simulations, multiple protein-cholesterol interaction points are observed and are related to known cholesterol recognition amino acid consensus and “inverted” cholesterol amino acid recognition consensus motifs.

For the PIP binding sites the authors identify a patch of four highly conserved lysines, ²¹⁶⁶KKKK²¹⁶⁹, at the cytoplasmic side of TM37 (Fig. 1 b), a Δ4K mutant of which had previously been shown to induce the xerocytosis channelopathy (9). In their electrophysiology studies, a Δ4K mutant showed an enhancement in ionic transport due to a reduction in channel inactivation, suggesting that PIP binding may have a role in regulating channel current by binding to this site. Similarly, simulations of this deletion also resulted in reduced binding to this site by PIPs. From simulations, they also highlight a second binding site for PIPs involving residues K2096, K2097, and K2163 (Fig. 1 b). However, in this case, a subsequent K2097E mutation had limited impact in both simulation and electrophysiology studies. A third binding site for PIPs involved R808, which causes xerocytosis when mutated to glutamine (10). Here, an R808Q mutant showed reduced PIP binding in simulations, but this was not further studied with electrophysiology experiments.

The channel also contains a wide assortment of proposed cholesterol recognition amino acid consensus (19) and cholesterol amino acid recognition consensus (39) binding motifs, many of which show cholesterol binding during the MD simulations. They mutate five of the residues associated with cholesterol binding, three of which are in the TM33-TM36 four-helix bundle, L1966A, L2037A, and Y2073A (a residue in the elbow helix), P2113A, and Y2173A, a residue adjacent to the KKKK motif in TM37. The two leucine mutants have limited impact on either pressure-induced current or inactivation. Y2073A appears to be nonfunctional and is suggested to be important for retaining the integrity of the four-helix bundle. Both P2113 and Y2173A impact upon pressure-induced activation but have limited effect on inactivation. The simulations show that the P2113A mutation reduces cholesterol binding, but how this relates to its role in pressure-activation is still to be determined.

This work shows the value of combining simulations with electrophysiological studies in identifying key and clinically relevant residues and elucidating their role in both normal channel function and their impact on the disease state. Here, large-scale, mixed membrane simulations enable the identification of binding sites through the probing of lipid interaction sites. This yields targets for mutagenesis studies, both computationally and through electrophysiological recordings. Ultimately, lipids are essential to the normal function of the PIEZO1 channel. The large-scale structural rearrangements involved in channel gating under pressure must be coupled to both the overall shape of the membrane and by lipids binding to tightly coordinated sites. This thereby provides a mechanism for the bilayer to modulate the rates of both activation and inactivation in the PIEZO1 mechanosensitive ion channel.

Future perspectives should address the energetics of binding of the identified lipids to the distinct sites and characterize the lipid selectivity for these sites, with both phosphatidylinositol-3,4-bisphosphate and PIP₃ competing for the same site in a number of instances. Ultimately a direct functional and/or structural role for both PIs and sterols in channel function should be sought, while also appreciating how glycolipids in the outer leaflet of the membrane influence channel curvature.

Editor: Philip Biggin.